Abstract
Background
In view of the established role of endothelial dysfunction in the pathophysiology of chronic obstructive pulmonary disease (COPD), the Endothelial Activation and Stress Index (EASIX), a validated biomarker of endothelial injury, warrants investigation as a potential prognostic tool. We therefore investigated its association with outcomes in critically ill COPD patients.
Methods
This retrospective cohort study analyzed data from critically ill patients with COPD in the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. The exposure was log2(EASIX) (continuous or tertiles). The outcomes were 28‑, 60‑, and 90‑day all‑cause mortality. Cox regression was primarily used to assess the association, and multiple additional approaches were employed to verify its robustness.
Results
A total of 1534 patients were included. Kaplan‑Meier survival curves showed significantly lower survival probabilities in higher EASIX tertiles (log‑rank p < 0.01 for all comparisons). Multivariable Cox regression confirmed that higher log2(EASIX) was independently associated with increased 28‑, 60‑, and 90‑day all‑cause mortality, with adjusted hazard ratios (HRs) of 1.87 (95% CI: 1.31–2.58), 1.83 (95% CI: 1.33–2.52), and 1.84 (95% CI: 1.36–2.50), respectively (all p < 0.01). Restricted cubic spline analysis indicated a linear relationship, and the association was robust across clinical subgroups. Sensitivity analyses in the fully imputed cohort confirmed linearity at 28 and 60 days but revealed a U‑shaped relationship at 90 days, with a nadir at log2(EASIX) = 0.56 (original EASIX = 1.47) and a HR of 0.74 (95% CI: 0.58–0.96). The E‑value of 2.96 (lower confidence bound 1.88) indicates robustness to unmeasured confounding.
Conclusion
Higher EASIX is independently associated with 28-day and 60-day mortality in critically ill COPD patients, whereas the 90-day relationship appears more complex. These findings suggest that EASIX may aid early risk assessment, pending external validation.
Keywords: endothelial activation and stress index, COPD, mortality, MIMIC-IV
Introduction
Chronic obstructive pulmonary disease (COPD) is a common chronic respiratory condition that imposes a substantial socioeconomic burden worldwide, and its prevalence is projected to increase by nearly 23% by 2050.1–3 Despite optimized management, critically ill COPD patients often require intensive care unit (ICU) admission and face an in‑hospital mortality rate as high as 20%-40%.4–6 Existing ICU prognostic scores, such as the Sequential Organ Failure Assessment (SOFA) and the Simplified Acute Physiology Score II (SAPS II), integrate multiple physiological and laboratory parameters and achieve relatively high predictive accuracy.7,8 However, these scores are often complex, not readily available in real time, and fail to directly capture endothelial dysfunction, which is a key driver of critical illness progression. Consequently, simple, rapid, and cost‑effective biomarkers are urgently needed for early risk stratification, particularly in resource‑limited settings.
The endothelial activation and stress index (EASIX), calculated from routinely measured serum lactate dehydrogenase (LDH), creatinine, and platelet (PLT) levels, is a readily available laboratory marker reflecting endothelial dysfunction and systemic inflammation.9 Originally developed to assess endothelial injury after hematopoietic stem cell transplantation,10 EASIX has subsequently been reported not only to be associated with mortality risk in sepsis, COVID‑19, and hypertension but also to serve as a prognostic biomarker in several respiratory diseases, including asthma, small cell lung cancer, and acute respiratory distress syndrome.11–16
A recent study using machine learning reported that EASIX predicts 28‑day mortality in critically ill COPD patients, but did not examine mid‑term outcomes at 60 and 90 days.17 Endothelial dysfunction is a pivotal driver of COPD progression and its common comorbidities, including cardiovascular diseases, and contributes to severe clinical outcomes.18–20 However, the association of EASIX with short- to mid-term mortality in critically ill COPD patients remains inadequately characterized. In this study, we position EASIX as a simple bedside risk marker that may complement rather than replace existing severity scores, offering added value in resource-limited settings while also providing mechanistic insight into endothelial involvement. To this end, we used the MIMIC-IV database to investigate the association between EASIX and 28-, 60-, and 90-day all-cause mortality in critically ill COPD patients.
Methods
Data Source
This study utilized data from the Medical Information Mart for Intensive Care IV (MIMIC-IV, version 3.0), a publicly available critical care database jointly developed by the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center. This database includes more than 90,000 patients admitted to the ICU of Beth Israel Deaconess Medical Center from 2001 to 2022.21 We obtained access to the database by passing the Protecting Human Research Participants course (No. 42460855). All data supporting the findings of this study are available from the MIMIC-IV database (https://mimic.physionet.org/).
Participant Selection Criteria
Patients with COPD who were admitted based on the International Classification of Diseases 9 codes (ICD-9) and ICD-10 codes (ICD-9:49,120, 49,121, 49,122, 496; ICD-10: J44, J440, J441, J449). All adult patients (age ≥18 years) hospitalized in the ICU for more than 24 hours were included in the study, and for patients who had multiple ICU admissions, only the first ICU stay data were used. Patients were excluded if they stayed in the ICU for less than 24 hours, or without unavailable measurements of LDH, creatinine, or platelets within 24 hours of admission. We extracted several relevant variables from the MIMIC IV database using Structured Query Language in Navicat Premium (version 11.2.7).
Study Variables and Outcomes
The age, sex, weight, height and ethnicity of the patients admitted to the ICU were collected. The scoring systems used to assess disease severity included SOFA, SAPS II, Logistic Organ Dysfunction Score (LODS), Physiology Score III (APS III), Model for End-Stage Liver Disease (MELD) and Glasgow Coma Scale (GCS). The comorbidities of sepsis, congestive heart failure, peripheral vascular disease, cerebrovascular disease, renal disease, liver disease and myocardial infarction were also collected. Vital signs included heart rate, mean blood pressure, respiratory rate, temperature and peripheral capillary oxygen saturation (SpO2). Laboratory parameters included white blood cell count (WBC), PLT, hemoglobin, neutrophils, lymphocytes, prothrombin time (PT), fibrinogen (FIB), international normalized ratio (INR), partial thromboplastin time (PTT), potential of hydrogen (pH), partial pressure of oxygen (PaO2), partial pressure of carbon dioxide (PaCO2), PaO2/FiO2 ratio, alanine aminotransferase (ALT), aspartate aminotransferase (AST), glucose, creatinine, blood urea nitrogen (BUN), albumin, LDH, total bilirubin (TBIL),sodium, chloride, anion gap, potassium, and calcium. The usages of mechanical ventilation, renal replacement therapy (RRT), vasopressor support were also collected. All study variables were measured within the first 24 hours of ICU admission, and for variables with a single measurement that value was used, while for those with multiple measurements the worst value (defined as the one most indicative of disease severity) was selected for analysis. EASIX was calculated by the formula: LDH (U/L)×creatinine (mg/dL) /PLT (109 cells/L). Multiple imputation was performed using the “MICE” package with 10 imputations (m = 10), assuming data were missing at random. For continuous variables, predictive mean matching was used; for binary variables, logistic regression was applied. The imputation model included all analysis variables. The proportion of missing values for each key variable is shown in Table S1.
The primary outcome was 28-day all-cause mortality after ICU admission. The secondary outcome was 60-day and 90-day all-cause mortality after ICU admission.
Statistical Analysis
If a continuous variable conformed to a normal distribution, it was expressed as mean±SD values, and otherwise as median and interquartile-range values. Categorical variables were described using frequencies and percentages.
Given the skewed distribution of EASIX, it was log2-transformed prior to analysis. Participants were then stratified into tertiles based on the transformed values. The Kaplan-Meier method was used to construct and compare 28-, 60-, and 90-day survival curves across EASIX tertiles, with differences assessed by the Log rank test. Cox proportional hazards models were used to evaluate the association between log2(EASIX) and mortality risk at 28, 60 and 90 days, thereby assessing its predictive performance for all-cause death. To ensure robust findings, we assessed the link between log2(EASIX) and all-cause mortality by constructing three sequentially adjusted Cox models (Model 1, Model 2, Model 3) in COPD patients. Model 1 was unadjusted; Model 2 was adjusted for demographic factors (age, sex, ethnicity, BMI); and Model 3 was further adjusted for a broad set of clinical scores (SOFA, SAPS II, LODS, APS III and GCS, comorbidities (sepsis, congestive heart failure, peripheral vascular disease, cerebrovascular disease, renal disease, liver disease and myocardial infarction), vital signs (heart rate, mean blood pressure, respiratory rate, temperature and SpO2), laboratory values (WBC, hemoglobin, neutrophils, lymphocytes, PT, INR, PTT, pH, PaO2, PaCO2, PaO2/FiO2 ratio, albumin, ALT, TBIL, glucose, BUN, sodium, chloride, anion gap, potassium, and calcium) and treatment (mechanical ventilation, RRT and vasopressor support). Multicollinearity among the severity scores was assessed using the variance inflation factor (VIF), and all VIF values were below 10, indicating no severe multicollinearity. The proportional hazards assumption was formally assessed using Schoenfeld residuals. Restricted cubic spline (RCS) analysis was used to examine the non-linear relationship between log2(EASIX) and 28-,60- and 90-day all-cause mortality in patients with COPD.
In addition, subgroup analyses by age, sex, SOFA, APS III, sepsis and mechanical ventilation were undertaken to assess the robustness of the associations. To minimize bias from excluding patients with incomplete laboratory data, we performed multiple imputation to re‑include all 4187 ICU COPD patients (including the 2653 previously excluded due to missing LDH, platelets, or creatinine) as a sensitivity analysis. Multiple imputation was performed using the “MICE” package in R with 10 imputations (m = 10). In this complete cohort, we repeated the Cox regression analysis with log2(EASIX) as a continuous variable and RCS analyses for 28‑day, 60‑day, and 90‑day all-cause mortality. E‑value was calculated to estimate the magnitude of an unmeasured confounding variable that would be required to mitigate the observed association between EASIX tertile (highest vs. lowest) and 28‑day mortality. The E‑value was derived from the hazard ratio (HR) and its 95% confidence interval (CI) obtained from the fully adjusted Cox proportional hazards model.22
IBM SPSS Statistics (version 26) and R software (version 4.5.1) were used for all statistical analyses. Multiple interpolation was used to fill in the missing values. Statistically significant differences were considered when p<0.05.
Results
Participant Baseline Characteristics
After applying the inclusion and exclusion criteria, 1534 COPD patients were included in the analysis (Figure 1); The baseline characteristics of the included cohort are summarized in Table 1. The median log2(EASIX) was 1.16 (interquartile range (IQR) 0.15–2.51). Participants were divided into three tertiles based on their log2(EASIX) values. The median (IQR) log2(EASIX) for Tertiles 1, 2, and 3 was −0.30 (−0.84 to 0.15), 1.17 (0.84 to 1.56), and 3.19 (2.51 to 4.48), respectively. Patients with higher log2(EASIX) were younger, more often female, and had higher BMI. They also had a higher prevalence of comorbidities including sepsis, congestive heart failure, peripheral vascular disease, renal disease, severe liver disease, and myocardial infarction, as well as higher SOFA, SAPS II, LODS, APS III and MELD scores, lower GCS score, and required more frequent vasopressor and RRT support. Vital signs showed that higher log2(EASIX) levels correlated with lower mean blood pressure, whereas SpO2 showed no significant linear correlation with log2(EASIX). Additionally, elevated log2(EASIX) was positively associated with increases in PT, INR, PTT, PaO2, PaO2/FiO2 ratio, ALT, LDH, TBIL, glucose, creatinine, BUN, chloride, anion gap and potassium, whereas it was inversely associated with platelets, hemoglobin, lymphocytes, albumin, sodium, and calcium levels. WBC and neutrophil counts were highest in the middle tertile of log2(EASIX), while pH and PaCO2 were lowest in the middle tertile. This study showed that higher log2(EASIX) was strongly associated with increased 28-day,60-day and 90-day all-cause mortality (all p < 0.01, Table 1) in COPD patients, whose overall mortality rates were 27.38%,33.12% and 35.92%, respectively. See Table 1 for detailed outcomes.
Figure 1.
Patient flow diagram.
Table 1.
Baseline Characteristics of the Study Population
| Variables | Overall (N=1534) | Tertile 1 (N = 512) | Tertile 2 (N = 511) | Tertile 3 (n = 511) | p |
|---|---|---|---|---|---|
| log2(EASIX) | 1.16 (0.15–2.51) | −0.30 (−0.84–0.15) | 1.17 (0.84–1.56) | 3.19 (2.51–4.48) | <0.01 |
| Age- years | 71. 81 (63.56–79.62) | 74.09 (65.84–81.23) | 70.89 (62.99–78.85) | 70.50 (62.28–78.44) | <0.01 |
| Sex- n (%) | <0.01 | ||||
| Men | 710 (46.28) | 269 (52.54) | 236 (46.18) | 205 (40.12) | |
| Female | 824 (53.72) | 243 (47.46) | 275 (53.82) | 306 (59.88) | |
| BMI | 27.60 (23.15–33.20) | 26.29 (22.06–31.74) | 27.52 (23.02–32.74) | 28.89 (24.60–34.73) | <0.01 |
| Ethnicity- n (%) | 0.21 | ||||
| Asian | 28 (1.83) | 9 (1.76) | 11 (2.15) | 8 (1.57) | |
| Black | 93 (6.06) | 23 (4.49) | 32 (6.26) | 38 (7.44) | |
| White | 1081 (70.47) | 382 (74.61) | 352 (68.88) | 347 (67.91) | |
| Other | 332 (21.64) | 98 (19.14) | 116 (22.70) | 118 (23.09) | |
| Vital signs | |||||
| Heart rate- beats/min | 86.96 (75.93–98.95) | 87.68 (77.44–99.50) | 86.41 (75.24–97.84) | 87.32 (75.55–100.10) | 0.28 |
| Respiratory rate- beats/min | 19.96 (17.52–22.62) | 19.94 (17.60–22.12) | 19.96 (17.69–22.90) | 20.04 (17.30–22.94) | 0.80 |
| SBP (mmHg) | 113.92 (104.71–125.28) | 116.32 (106.94–128.05) | 113.95 (104.54–125.54) | 111.62 (103.00–122.12) | <0.01 |
| DBP (mmHg) | 60.58 (54.70–67.58) | 62.55 (56.34–69.34) | 59.96 (53.74–67.29) | 59.68 (54.02–66.33) | <0.01 |
| MAP (mmHg) | 75.54 (69.12–82.43) | 76.58 (70.49–84.01) | 74.76 (68.96–82.39) | 74.74 (68.04–81.23) | <0.01 |
| Temperature (°C) | 36.79 (36.52–37.06) | 36.79 (36.57–37.03) | 36.82 (36.53–37.09) | 36.75 (36.44–37.09) | 0.35 |
| SPO2 (%) | 96.25 (94.54–97.76) | 95.89 (94.27–97.57) | 96.50 (94.70–97.86) | 96.23 (94.64–97.85) | 0.02 |
| Comorbidities- n (%) | |||||
| Sepsis | 1098 (71.58) | 286 (55.86) | 391 (76.52) | 421 (82.39) | <0.01 |
| Congestive heart failure | 738 (48.11) | 194 (37.89) | 274 (53.62) | 270 (52.84) | <0.01 |
| Peripheral vascular disease | 280 (18.25) | 65 (12.70) | 99 (19.37) | 116 (22.70) | <0.01 |
| Cerebrovascular disease | 168 (10.95) | 49 (9.57) | 58 (11.35) | 61 (11.94) | 0.45 |
| Renal disease | 393 (25.62) | 50 (9.77) | 137 (26.81) | 206 (40.31) | <0.01 |
| Severe liver disease | 78 (5.08) | 8 (1.56) | 12 (2.35) | 58 (11.35) | <0.01 |
| Myocardial infarction | 370 (24.12) | 81 (15.82) | 132 (25.83) | 157 (30.72) | <0.01 |
| Laboratory parameters | |||||
| WBC (k/ul) | 13.05 (9.10–17.90) | 12.15 (9.00–16.50) | 13.70 (9.50–18.80) | 13.10 (8.90–18.70) | 0.01 |
| PLT (k/ul) | 180.00 (122.00–246.00) | 238.50 (189.75–324.50) | 174.00 (132.00–225.00) | 117.00 (74.50–169.50) | <0.01 |
| Hemoglobin (g/dL) | 11.20 (9.80–12.80) | 11.30 (9.90–12.90) | 11.40 (10.00–12.80) | 10.90 (9.50–12.40) | <0.01 |
| Neutrophils (k/ul) | 9.97 (6.72–14.23) | 9.62 (6.81–13.33) | 10.58 (6.95–15.63) | 9.77 (6.04–14.35) | 0.03 |
| Lymphocytes (k/ul) | 0.78 (0.45–1.27) | 0.87 (0.52–1.36) | 0.77 (0.47–1.31) | 0.72 (0.40–1.18) | <0.01 |
| PT (second) | 14.50 (12.80–18.00) | 13.50 (12.10–15.80) | 14.50 (12.95–17.70) | 15.90 (13.50–20.90) | <0.01 |
| INR | 1.30 (1.10–1.70) | 1.20 (1.10–1.40) | 1.30 (1.20–1.60) | 1.40 (1.20–1.90) | <0.01 |
| PTT (second) | 33.60 (28.80–47.63) | 31.60 (27.60–37.43) | 33.20 (28.60–44.85) | 38.40 (30.55–59.75) | <0.01 |
| PH | 7.30 (7.21–7.37) | 7.31 (7.21–7.38) | 7.28 (7.19–7.36) | 7.31 (7.23–7.37) | <0.01 |
| PaO2 (mmHg) | 60.00 (40.00–83.00) | 57.50 (38.00–80.00) | 60.00 (39.00–81.00) | 65.00 (43.00–89.00) | 0.03 |
| PaCO2 (mmHg) | 52.00 (43.00–65.00) | 52.00 (43.00–66.00) | 50.00 (43.00–63.25) | 53.00 (44.00–65.50) | 0.28 |
| PaO2/FiO2 ratio | 122.00 (78.33–195.00) | 117.33 (74.71–198.75) | 116.75 (75.00–185.75) | 132.00 (89.00–213.00) | 0.01 |
| ALB (g/dL) | 3.20 (2.70–3.60) | 3.30 (2.80–3.60) | 3.20 (2.70–3.60) | 3.00 (2.60–3.50) | <0.01 |
| ALT (U/L) | 27.00 (17.00–63.00) | 21.50 (15.00–40.00) | 27.00 (16.00–61.50) | 38.00 (20.00–128.50) | <0.01 |
| LDH (IU/L) | 293.00 (214.00–453.75) | 217.50 (176.75–279.00) | 289.00 (230.00–388.00) | 488.00 (315.50–969.50) | <0.01 |
| TBIL (mg/dL) | 0.60 (0.40–1.20) | 0.50 (0.30–0.80) | 0.70 (0.40–1.10) | 0.90 (0.50–2.00) | <0.01 |
| Glucose (mg/dL) | 135.69 (113.00–171.00) | 129.42 (109.00–161.54) | 137.80 (116.64–171.17) | 138.75 (114.69–178.10) | 0.00 |
| Cr (mg/dL) | 1.20 (0.80–1.90) | 0.80 (0.60–1.10) | 1.30 (1.00–1.70) | 2.10 (1.40–3.30) | <0.01 |
| BUN (mg/dL) | 29.00 (19.00–46.00) | 20.00 (14.00–28.00) | 29.00 (20.50–43.00) | 44.00 (29.00–65.00) | <0.01 |
| Sodium (mEq/L) | 137.00 (134.00–140.00) | 138.00 (134.00–141.00) | 137.00 (134.00–140.00) | 137.00 (133.00–140.00) | 0.01 |
| Chloride (mEq/L) | 105.00 (100.00–109.00) | 103.00 (99.00–107.00) | 105.00 (101.00–110.00) | 105.00 (101.00–110.00) | <0.01 |
| Anion gap (mmol/L) | 16.00 (13.00–19.00) | 14.00 (12.00–17.00) | 16.00 (13.00–18.00) | 18.00 (15.00–21.00) | <0.01 |
| Potassium (mEq/L) | 3.90 (3.60–4.40) | 3.90 (3.60–4.30) | 3.90 (3.60–4.30) | 4.00 (3.60–4.50) | 0.01 |
| Calcium (mEq/L) | 8.10 (7.50–8.68) | 8.40 (7.90–8.90) | 8.10 (7.50–8.50) | 7.70 (7.20–8.50) | <0.01 |
| Score | |||||
| SOFA | 5.00 (3.00–8.00) | 3.00 (1.00–5.00) | 5.00 (3.00–7.00) | 8.00 (6.00–11.00) | <0.01 |
| SAPS II | 40.00 (32.00–51.00) | 35.00 (28.00–41.00) | 42.00 (34.00–49.00) | 49.00 (39.00–59.00) | <0.01 |
| LODS | 5.00 (3.00–8.00) | 4.00 (2.00–6.00) | 5.00 (4.00–7.00) | 7.00 (5.00–9.00) | <0.01 |
| APS III | 48.00 (37.00–64.00) | 40.500 (32.00–50.00) | 48.00 (37.50–60.00) | 60.00 (47.00–78.00) | <0.01 |
| GCS | 15.00 (14.00–15.00) | 15.00 (14.00–15.00) | 15.00 (14.00–15.00) | 15.00 (13.00–15.00) | <0.01 |
| Treatments- n (%) | |||||
| Mechanical ventilation | 606 (39.50) | 214 (41.80) | 194 (37.96) | 198 (38.75) | 0.42 |
| RRT | 39 (2.54) | 0 (0.00) | 1 (0.20) | 38 (7.44) | <0.01 |
| Vasopressor support | 667 (43.48) | 123 (24.02) | 248 (48.53) | 296 (57.93) | <0.01 |
| Clinical outcomes- n (%) | |||||
| 28-Day-mortality | 420 (27.38) | 84 (16.41) | 135 (26.42) | 201 (39.33) | <0.01 |
| 60-Day mortality | 508 (33.12) | 114 (22.27) | 166 (32.49) | 228 (44.62) | <0.01 |
| 90-Day mortality | 551 (35.92) | 128 (25.00) | 182 (35.62) | 241 (47.16) | <0.01 |
Abbreviations: EASIX, Endothelial activation and stress index; HRs, Hazard ratios; LDH, Lactate dehydrogenase; PLT, Platelet; WBC, White blood cell; RBC, Red blood cell; INR, International normalized ratio; PT, Prothrombin time; PTT, Partial thromboplastin time; FIB, fibrinogen; pH, potential of hydrogen, PaO2, partial pressure of oxygen; PaCO2, partial pressure of carbon dioxide; ALT, Alanine aminotransferase; AST, Aspartate aminotransferase; BUN, Blood urea nitrogen; TBIL, Total bilirubin; SOFA, Sequential organ failure assessment; SAPS II, Simplified acute physiology score II; APS III, Acute physiology score III; GCS, Glasgow coma scale; LODS, Logistic Organ Dysfunction Score; MELD, Model for End-Stage Liver Disease; SpO2, peripheral capillary oxygen saturation; RRT, renal replacement therapy.
Association of log2(EASIX) with Survival Outcomes
The Kaplan-Meier curves in Figure 2 demonstrate significantly worse 28-day,60-day and 90-day survival for patients in the higher log2(EASIX) tertiles (log-rank p < 0.01 for all time points). In the highest tertile (Tertile 3), the number of survivors dropped from 511 at baseline to 311 at 28 days. Concomitantly, the lowest tertile (Tertile 1) consistently demonstrated the most favorable outcomes, with survivor counts ranging from 512 to 431 at 28 days. The findings were consistent at 60 and 90 days.
Figure 2.
Kaplan-Meier Survival Curves for All-Cause Mortality at 28-day (A) 60-day (B) and 90-day (C) Stratified by log2(EASIX) Tertiles (log-rank p < 0.01 for all comparisons).
The Cox proportional hazards models demonstrated that higher log2(EASIX) values were progressively associated with increased mortality risk across all follow-up periods, a finding that held for both continuous and categorical analyses of log2(EASIX) (Table 2). Each one-unit increase in log2(EASIX), which corresponds to a doubling of the original EASIX, was associated with a 26% higher risk of 28-day mortality, a 23% higher risk of 60-day mortality, and a 21% higher risk of 90-day mortality in unadjusted analyses. After full adjustment, the associations, although attenuated, remained significant, with corresponding risk increases of 22%, 21%, and 20%, respectively (all p < 0.01). Stratification by EASIX tertiles revealed a dose-dependent mortality risk. Compared to Tertile 1, the highest tertile (Tertile 3) had significantly elevated risks, with unadjusted HRs (95% CI) of 2.84 (2.20–3.67), 2.44 (1.95–3.06), and 2.32 (1.88–2.88) for 28-day, 60-day, and 90-day mortality, respectively. After full adjustment, the corresponding HRs were 1.87 (1.31–2.58), 1.83 (1.33–2.52), and 1.84 (1.36–2.50) (all p <0.01, Table 2). Assessment of the proportional hazard assumption for the primary Cox model using Schoenfeld residuals showed no evidence of violation, with a global p-value of 0.09 as shown in Figure S1.
Table 2.
Multivariable Cox Regression Analysis of log2(EASIX) and 28-Day,60-Day and 90-Day Mortality in Patients
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| HR (95% CI) | p | HR (95% CI) | p | HR (95% CI) | p | |
| 28-Day mortality | ||||||
| Log2(EASIX) | 1.26(1.21–1.31) | <0.01 | 1.29(1.24–1.35) | <0.01 | 1.22(1.15–1.32) | <0.01 |
| Tertile 1 | Ref. | Ref. | Ref. | |||
| Tertile 2 | 1.73(1.32–2.28) | <0.01 | 1.65(1.25–2.17) | <0.01 | 1.43(1.06–1.93) | 0.02 |
| Tertile 3 | 2.84(2.20–3.67) | <0.01 | 3.20(2.47–4.14) | <0.01 | 1.87(1.31–2.58) | <0.01 |
| 60-Day mortality | ||||||
| Log2(EASIX) | 1.23(1.18–1.27) | <0.01 | 1.26(1.21–1.31) | <0.01 | 1.21(1.14–1.30) | <0.01 |
| Tertile 1 | Ref. | Ref. | Ref. | |||
| Tertile 2 | 1.59(1.25–2.02) | <0.01 | 1.51(1.19–1.92) | <0.01 | 1.35(1.04–1.76) | 0.02 |
| Tertile 3 | 2.44(1.95–3.06) | <0.01 | 2.77(2.20–3.49) | <0.01 | 1.83(1.33–2.52) | <0.01 |
| 90-Day mortality | ||||||
| Log2(EASIX) | 1.21(1.17–1.26) | <0.01 | 1.24(1.19–1.29) | <0.01 | 1.20(1.13–1.28) | <0.01 |
| Tertile 1 | Ref. | Ref. | Ref. | |||
| Tertile 2 | 1.56(1.24–1.96) | <0.01 | 1.48(1.18–1.86) | <0.01 | 1.36(1.06–1.75) | 0.01 |
| Tertile 3 | 2.32(1.88–2.88) | <0.01 | 2.60(2.09–3.23) | <0.01 | 1.84(1.36–2.50) | <0.01 |
Notes: Model 1: unadjusted model. Model 2: adjusted for age, sex, BMI and ethnicity. Model 3: further adjusted for an extensive set of covariates beyond Model 2, including: clinical scores (SOFA, SAPS II, LODS, APS III and GCS), comorbidities (sepsis, congestive heart failure, peripheral vascular disease, cerebrovascular disease, renal disease, liver disease and myocardial infarction), vital signs (heart rate, mean blood pressure, respiratory rate, temperature and SpO2), laboratory values(WBC, hemoglobin, neutrophils, lymphocytes, PT, INR, PTT, pH, PaO2, PaCO2, PaO2/FiO2 ratio, albumin, ALT, TBIL, glucose, BUN, sodium, chloride, anion gap, potassium, and calcium) and treatment (mechanical ventilation, RRT and vasopressor support).
Abbreviations: HR, hazard ratio; CI, confidence interval; EASIX, Endothelial Activation and Stress Index.
We applied RCS analysis to assess the potential nonlinear association of log2(EASIX) with mortality in critically ill COPD patients. As shown in Figure 3, the relationship was predominantly linear at all follow-up times. Formal nonlinearity tests were non-significant (p-values: 0.66 at 28 days, and 0.41, 0.14 at subsequent time points). This supports a consistent linear association where each unit increase in log2(EASIX) corresponds to elevated mortality risk.
Figure 3.
Restricted cubic spline analysis for the association of log2(EASIX) with 28- (A), 60- (B), and 90-day (C) all-cause mortality in patients with COPD. 28-day all-cause mortality: p for no-linear = 0.66. 60-day all-cause mortality: p for no-linear = 0.41. 90-day all-cause mortality: p for no-linear = 0.14.
Subgroup Analyses
Subgroup analyses demonstrated a consistent association between higher log2(EASIX) and increased 28-day all-cause mortality across most strata, including sex, SOFA, APS III, sepsis, and mechanical ventilation, with no significant interactions (all p for interaction > 0.05) (Figure 4). In the age-stratified analysis, the association did not reach statistical significance in younger patients; however, the interaction test for age was also non-significant (p for interaction > 0.05), indicating that age does not modify the effect. Similar associations were observed for 60‑day and 90‑day all-cause mortality (Figure S2), further supporting the robustness of the findings.
Figure 4.
Subgroup forest plot for the association of log2(EASIX) with 28-day mortality in COPD patients.
Abbreviations: EASIX, Endothelial Activation and Stress Index; HR, hazard ratio; CI, confidence interval; Apsiii, Acute Physiology Score III; SOFA, Sequential Organ Failure Assessment.
Sensitivity Analyses
To further assess the robustness of our findings while minimizing selection bias due to incomplete data, we performed multiple imputation for all analysis variables in the 2653 patients who had been excluded because of missing LDH, platelets, or creatinine. After imputation, all 4187 ICU COPD patients were re‑included in the analysis. A comparison between the included and excluded patients showed that the excluded patients had significantly milder disease (Table S2). After imputation, the distribution of log2(EASIX) tertiles in the full cohort shifted toward lower values (Figure S3). This approach allowed us to evaluate the robustness of our results using complete data.
In this full cohort, the Cox model with log2(EASIX) as a continuous variable yielded hazard ratios for 28‑day, 60‑day, and 90‑day all-cause mortality that were consistent with the primary analysis (Table S3). Restricted cubic spline analyses showed a linear relationship for 28‑day and 60‑day all‑cause mortality (p for non‑linearity = 0.09 and 0.14, respectively), but a non‑linear relationship for 90‑day all‑cause mortality (p for non‑linearity < 0.01; Figure S4). The nadir of the U‑shaped curve was located at log2(EASIX) = 0.56 (original EASIX = 1.47), with a hazard ratio of 0.74 (95% CI: 0.58–0.96).
To further assess robustness to unmeasured confounding, we calculated E-values for the primary comparison (highest versus lowest EASIX tertile) for 28-day all-cause mortality. In addition, the resulting E-value for the point estimate was 2.96, and the E-value for the lower confidence bound was 1.88. This indicates that an unmeasured confounder would need to be associated with both EASIX tertile and 28-day all-cause mortality by a risk ratio of at least 2.96 (or 1.88 when considering the CI limit) to nullify the observed association, suggesting that our results are reasonably robust to unmeasured confounding.
Discussion
In this large MIMIC-IV cohort, higher EASIX values were associated with increased short- to mid-term all-cause mortality at 28, 60, and 90 days in critically ill COPD patients, and this association remained robust after adjustment for potential confounders and across key clinical subgroups, including age, sex, SOFA, APS III, sepsis, and mechanical ventilation. In our sensitivity analysis that re-included milder patients, the association remained robust for 28-day and 60-day all-cause mortality, but a non-linear U-shaped relationship was observed at 90 days, suggesting that the association between EASIX and longer-term mortality may be more complex and warrants further investigation.These findings suggest a potential role of endothelial dysfunction in the observed association between EASIX and mortality in critically ill COPD patients.
These results may be related to the involvement of endothelial dysfunction in the pathogenesis of COPD. Apoptosis of alveolar endothelial cells contributes to the development of COPD and emphysema.23 As early as 1998, Wiebe BM and Laursen H24 found that, compared with healthy individuals, patients with COPD exhibited reduced capillary length and density, primarily caused by endothelial cell apoptosis and endothelial dysfunction. Impaired vasodilatory function of the vascular endothelium promotes the infiltration of inflammatory cells into the small airways and leads to the accumulation of extracellular matrix, thereby exacerbating small airway remodeling and worsening the disease. Recent studies have shown that endothelial cells release endothelial microparticles from their surface upon activation or apoptosis, which play significant biological roles in both endothelial dysfunction and thrombosis.25,26 Animal experiments have further revealed that circulating endothelial microparticles released from apoptotic pulmonary capillary endothelial cells are associated with a decline in pulmonary function in rats exposed to cigarette smoke.27 At the clinical level, study has found that the number of circulating endothelial microparticles is significantly increased in patients during acute exacerbation of COPD,23 directly indicating that endothelial dysfunction is associated with the severity of COPD. Moreover, experiments have demonstrated that transplantation of healthy endothelial cells can repair alveolar damage induced by elastase, thereby improving lung elasticity and respiratory function.28 Therefore, compelling evidence supports the use of endothelial dysfunction markers as biomarkers for COPD.
The three components of the EASIX score, namely LDH, creatinine, and platelets, have each been independently validated as predictors of clinical outcomes in COPD patients. LDH reflects cellular damage,29,30 creatinine indicates renal impairment which is often exacerbated by hypercapnia or oxidative stress,31–33 and platelets are involved in systemic inflammation, microvascular injury, and airway remodeling.34,35 Therefore, it is plausible that the association between EASIX and mortality is partly explained by endothelial dysfunction and overall illness severity in critically ill COPD patients.
Our sensitivity analysis showed that the positive association between higher EASIX and 28‑day and 60‑day mortality remained robust, whereas a U‑shaped pattern was observed only at 90 days. This time‑dependent pattern may be explained by the changing causes of death over time. In the acute phase (28‑60 days), mortality is primarily driven by acute organ failure directly related to endothelial dysfunction and microcirculatory injury.36,37 This is consistent with the natural history of COPD, where mortality risk peaks within the first week after admission and declines sharply by three months.38 Our study also supports this, as patients in the highest EASIX tertile had higher severity scores and required more frequent organ support. This may explain why higher EASIX is associated with increased 28‑day and 60‑day mortality. Therefore, EASIX may serve as a practical complement to existing severity scores, particularly for early risk assessment in critically ill COPD patients.
By 90 days, however, the composition of deaths shifts. First, a substantial proportion of high‑risk patients (those with the highest EASIX values) have already died within the first 28 to 60 days. This means the surviving cohort at 90 days is selectively enriched with lower‑risk individuals, a phenomenon known as survivor bias. Moreover, the sensitivity analysis cohort was older, which may lead to a higher proportion of non‑acute events such as COPD progression, cardiovascular events, and acute exacerbations at this later time point.38,39 Second, the sensitivity analysis cohort exhibited a left‑shifted distribution of EASIX, introducing a substantial number of patients with very low EASIX values. These very low values may serve as markers of malnutrition or frailty. For example, very low creatinine may reflect reduced muscle mass, very low platelets may indicate bone marrow suppression, and very low LDH may suggest a hypo‑metabolic state. These frail patients survived the acute phase but become vulnerable to late non‑acute deaths.40,41 This phenomenon was not apparent at 28 or 60 days because acute organ failure dominated the early phase, masking the effects of frailty. Given these findings, we acknowledge the possibility that EASIX may, at least in part, function as a global marker of critical illness rather than exclusively reflecting COPD‑specific endothelial dysfunction.
In our study, even after adjustment for traditional severity scores, EASIX remained independently associated with mortality, indicating that it provides additional risk information beyond what can be captured by SOFA or SAPS II alone. Notably, EASIX integrates markers of tissue damage (LDH), renal perfusion (creatinine), and thrombo‑inflammation (PLT), which may offer particular added value in initial risk assessment. Thus, in resource‑limited settings where detailed severity scores are difficult to obtain, EASIX could serve as a simplified but valuable alternative for initial risk stratification. Conversely, in well‑equipped ICUs, EASIX could complement existing scores as a dynamic adjunct, improving the granularity of risk assessment without increasing complexity. Thus, EASIX may be most valuable for early risk identification and intervention, which is the primary clinical need in critically ill COPD patients.
Limitations
However, this study has several limitations. First, our findings are derived from a single database (MIMIC‑IV), which may limit their generalizability to other ICU settings or patient populations with different case mixes, healthcare systems, or resource availability. Second, despite performing multiple imputation followed by sensitivity analyses and calculating relatively high E‑values, potential selection bias cannot be completely ruled out. Third, since our research primarily focused on exploring the potential of EASIX for early identification of COPD patients with poor prognosis, we only assessed EASIX at the time of ICU admission. External prospective validation in a multicenter cohort is required to confirm the generalizability of our findings across different ICU settings and populations.
Conclusion
Our study demonstrated that elevated EASIX is independently associated with 28‑day and 60‑day mortality in critically ill COPD patients, whereas the 90‑day relationship is more complex. These findings suggest that EASIX may aid early risk assessment, particularly within the first two months, pending external validation in independent cohorts.
Acknowledgments
We thank the Massachusetts Institute of Technology and the Beth Israel Deaconess Medical Center for providing the MIMIC-IV database.
Funding Statement
This study were supported by the Science and Technology Projects in Guangzhou (grant number 2024A03J1035) and Guangzhou Science and Technology Plan Project 2024 Basic and Applied Basic Research Special Topic Young Doctoral Start-Up Project (grant number 2024A04J5142).
Data Sharing Statement
Data supporting the findings of this study were derived from the MIMIC-IV database (https://mimic.physionet.org/).
Ethics Approval and Consent to Participate
The data used in this study are from the MIMIC database, a publicly available de-identified database. The original database received ethical approval from the Institutional Review Board of the Massachusetts Institute of Technology (MIT) and Beth Israel Deaconess Medical Center (BIDMC) and obtained patient informed consent. Our institutional ethics committee reviewed this study and confirmed that, according to the Ministry of Health’s “Measures for Ethical Review of Biomedical Research Involving Human Subjects (Trial)” and Article 32 (1)(2) of China’s “Measures for Ethical Review of Life Science and Medical Research Involving Human Subjects” (2023), no further ethical review or filing is required for the use of this public anonymous database.
Author Contributions
All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.
Disclosure
The authors declare that they have no competing interests in this work.
References
- 1.Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Pulmonary Disease (2025 report). Global initiative for chronic obstructive lung disease.
- 2.Viegi G, Pistelli F, Sherrill DL, et al. Definition, epidemiology and natural history of COPD. Eur Respir J. 2007;30(5):993–12. doi: 10.1183/09031936.00082507 [DOI] [PubMed] [Google Scholar]
- 3.Boers E, Barrett M, Su JG, et al. Global burden of chronic obstructive pulmonary disease through 2050. JAMA Network Open. 2023;6(12):e2346598. doi: 10.1001/jamanetworkopen.2023.46598 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ai L, Zhang J, Bo W. Respiratory failure in intensive care unit patients with progressive COPD: nursing approaches to patient care. Altern Ther Health Med. 2022;28(1):52–57. [PubMed] [Google Scholar]
- 5.Vlachaki I, Donhauser S, Wise RA, et al. A systematic literature review of the humanistic, economic, sociodemographic, and environmental burden associated with severe COPD. Int J Chron Obstruct Pulmon Dis. 2025;20:2493–2523. doi: 10.2147/COPD.S510623 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Prediletto I, Giancotti G, Nava S. COPD exacerbation: why it is important to avoid ICU admission. J Clin Med. 2023;12(10):3369. doi: 10.3390/jcm12103369 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Vincent JL, de Mendonça A, Cantraine F, et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Working group on “sepsis-related problems” of the European Society of Intensive Care Medicine. Crit Care Med. 1998;26(11):1793–1800. doi: 10.1097/00003246-199811000-00016 [DOI] [PubMed] [Google Scholar]
- 8.Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA. 1993;270(24):2957–2963. Erratum in: JAMA 1994 May 4;271(17):1321. doi: 10.1001/jama.270.24.2957 [DOI] [PubMed] [Google Scholar]
- 9.Song GY, Jung SH, Kim K, et al. Endothelial activation and stress index (EASIX) is a reliable predictor for overall survival in patients with multiple myeloma. BMC Cancer. 2020;20(1):803. doi: 10.1186/s12885-020-07317-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Luft T, Benner A, Jodele S, et al. EASIX in patients with acute graft-versus-host disease: a retrospective cohort analysis. Lancet Haematol. 2017;4(9):e414–e423. doi: 10.1016/S2352-3026(17)30108-4 [DOI] [PubMed] [Google Scholar]
- 11.Korell F, DeFilipp Z, Schreck N, et al. Taskforces allogeneic stem cell transplantation. validation of pre-conditioning EASIX for prediction of sepsis after allogeneic stem cell transplantation. Intensive Care Med. 2023;49(11):1408–1410. doi: 10.1007/s00134-023-07193-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kalicińska E, Biernat M, Rybka J, et al. Endothelial activation and stress index (EASIX) as an early predictor for mortality and overall survival in hematological and non-hematological patients with COVID-19: multicenter cohort study. J Clin Med. 2021;10:4373. doi: 10.3390/jcm10194373 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Murat E, Karpat MS, Öztürk Y, et al. The role of the EASIX score in patients with hypertension: a cross-sectional study. Egypt Heart J. 2025;77(1):112. doi: 10.1186/s43044-025-00710-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.He Y, Li Y, Xiaojin Y, et al. Endothelial activation and stress index for prediction of mortality in asthma. Front Med. 2025;12:1622944. doi: 10.3389/fmed.2025.1622944 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Chen J, Lv J, Liu M, et al. Between endothelial activation and stress index (EASIX) and mortality in acute respiratory distress syndrome (ARDS) patients: a multicenter retrospective study. Front Physiol. 2025;16:1570988. doi: 10.3389/fphys.2025.1570988 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Go SI, Park S, Kang MH, et al. Endothelial activation and stress index (EASIX) as a predictive biomarker in small cell lung cancer. Cancer Biomark. 2022;35(2):217–225. doi: 10.3233/CBM-220032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Niu J, Huang Q, Dong Y, et al. Association between the endothelial activation and stress index and 28-day all-cause mortality in critically ill patients with chronic obstructive pulmonary disease: a retrospective cohort study and predictive model establishment based on machine learning. BMC Pulm Med. 2026. doi: 10.1186/s12890-026-04299-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kovacs G, Agusti A, Barberà JA, et al. Pulmonary vascular involvement in chronic obstructive pulmonary disease. Is there a pulmonary vascular phenotype? Am. J. Respir Crit. Care Med. 2018;198:1000–1011. doi: 10.1164/rccm.201801-0095PP [DOI] [PubMed] [Google Scholar]
- 19.Theodorakopoulou MP, Alexandrou ME, Bakaloudi DR, et al. Endothelial dysfunction in COPD: a systematic review and meta-analysis of studies using different functional assessment methods. E R J. Open. Res. 2021;7(2):00983–2020. doi: 10.1183/23120541.00983-2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Polverino F, Celli BR, Owen CA. COPD as an endothelial disorder: endothelial injury linking lesions in the lungs and other organs? (2017 grover conference series). Pulm Circ. 2018;8(1):2045894018758528. doi: 10.1177/2045894018758528 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Johnson A, Bulgarelli L, Shen L, et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci Data. 2023;10(1):1. doi: 10.1038/s41597-022-01899-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-Value. Ann Intern Med. 2017;167(4):268–274. doi: 10.7326/M16-2607 [DOI] [PubMed] [Google Scholar]
- 23.Takahashi T, Kobayashi S, Fujino N, et al. Increased circulating endothelial microparticles in COPD patients: a potential biomarker for COPD exacerbation susceptibility. Thorax. 2012;67(12):1067–1074. doi: 10.1136/thoraxjnl-2011-201395 [DOI] [PubMed] [Google Scholar]
- 24.Wiebe BM, Laursen H. Lung morphometry by unbiased methods in emphysema: bronchial and blood vessel volume, alveolar surface area and capillary length. APMIS. 1998;106(6):651–656. doi: 10.1111/j.1699-0463.1998.tb01395.x [DOI] [PubMed] [Google Scholar]
- 25.Combes V, Simon AC, Grau GE, et al. In vitro generation of endothelial microparticles and possible prothrombotic activity in patients with lupus anticoagulant. J Clin Invest. 1999;104(1):93–102. doi: 10.1172/JCI4985 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Trappe A, Donnelly SC, McNally P, et al. Role of extracellular vesicles in chronic lung disease. Thorax. 2021;76(10):1047–1056. doi: 10.1136/thoraxjnl-2020-216370 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Liu H, Ding L, Zhang Y, et al. Circulating endothelial microparticles involved in lung function decline in a rat exposed in cigarette smoke maybe from apoptotic pulmonary capillary endothelial cells. J Thorac Dis. 2014;6(6):649–655. doi: 10.3978/j.issn.2072-1439.2014.06.26 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hisata S, Racanelli AC, Kermani P, et al. Reversal of emphysema by restoration of pulmonary endothelial cells. J Exp Med. 2021;218(8):e20200938. doi: 10.1084/jem.20200938 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Papadopoulos D, Skopas V, Trakas N, et al. Serum lactate dehydrogenase and its isoenzymes as predictors of clinical outcomes in acute exacerbation of chronic obstructive pulmonary disease: a retrospective analysis of a hospitalized cohort. Monaldi Arch Chest Dis. 2023;94(2). doi: 10.4081/monaldi.2023.2543 [DOI] [PubMed] [Google Scholar]
- 30.Skopas V, Papadopoulos D, Trakas N, et al. Lactate dehydrogenase isoenzymes in patients with acute exacerbation of chronic obstructive pulmonary disease: an exploratory cross-sectional study. Respir Physiol Neurobiol. 2021;283:103562. doi: 10.1016/j.resp.2020.103562 [DOI] [PubMed] [Google Scholar]
- 31.Madouros N, Jarvis S, Saleem A, et al. Is there an association between chronic obstructive pulmonary disease and chronic renal failure? Cureus. 2022;14(6):e26149. doi: 10.7759/cureus.26149 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Chen W, FitzGerald JM, Sin DD, et al. Canadian respiratory research network. excess economic burden of comorbidities in COPD: a 15-year population-based study. Eur Respir J. 2017;50(1):1700393. doi: 10.1183/13993003.00393-2017 [DOI] [PubMed] [Google Scholar]
- 33.Levey AS, Eckardt KU, Tsukamoto Y, et al. Definition and classification of chronic kidney disease: a position statement from Kidney Disease: improving Global Outcomes (KDIGO). Kidney Int. 2005;67(6):2089–2100. doi: 10.1111/j.1523-1755.2005.00365.x [DOI] [PubMed] [Google Scholar]
- 34.Bozinovski S, Vlahos R, Anthony D, et al. COPD and squamous cell lung cancer: aberrant inflammation and immunity is the common link. Br J Pharmacol. 2016;173(4):635–648. doi: 10.1111/bph.13198 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Faner R, Tal-Singer R, Riley JH, et al. ECLIPSE Study Investigators. Lessons from ECLIPSE: a review of COPD biomarkers. Thorax. 2014;69(7):666–672. doi: 10.1136/thoraxjnl-2013-204778 [DOI] [PubMed] [Google Scholar]
- 36.Schønemann-Lund M, Itenov TS, Larsson JE, et al. Endotheliopathy is associated with slower liberation from mechanical ventilation: a cohort study. Crit Care. 2022;26(1):33. doi: 10.1186/s13054-021-03877-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Chi C, Zhou J, Hou S. Impact of early acute kidney injury on 30-day mortality in intensive care unit patients with chronic obstructive pulmonary disease: a retrospective study using MIMIC-IV. J Thorac Dis. 2025;17(9):6404–6415. doi: 10.21037/jtd-2025-18 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Suissa S, Dell’Aniello S, Ernst P. Long-term natural history of chronic obstructive pulmonary disease: severe exacerbations and mortality. Thorax. 2012;67(11):957–963. doi: 10.1136/thoraxjnl-2011-201518 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Fabbri LM, Celli BR, Agustí A, et al. COPD and multimorbidity: recognising and addressing a syndemic occurrence. Lancet Respir Med. 2023;11(11):1020–1034. doi: 10.1016/S2213-2600(23)00261-8 [DOI] [PubMed] [Google Scholar]
- 40.Ribeiro Baptista B, Valentin S, Chaouat A. Frailty in COPD: early presence, future consequences. Respir Med Res. 2026;90:101280. doi: 10.1016/j.resmer.2026.101280 [DOI] [PubMed] [Google Scholar]
- 41.Nishimura K, Kusunose M, Sanda R, et al. Frailty and exacerbation of chronic obstructive pulmonary disease: is there any association? Int J Chron Obstruct Pulmon Dis. 2024;19:1131–1139. doi: 10.2147/COPD.S455316 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data supporting the findings of this study were derived from the MIMIC-IV database (https://mimic.physionet.org/).