Blood urea nitrogen to albumin ratio and mortality in critically ill patients with cardiogenic shock Xu Zhu, Qiong Wu, Zuo Kang, Shu Chen, Zhengdong Hu ABSTRACT Purpose: Given the high mortality associated with cardiogenic shock (CS), this study aimed to evaluate the association between the blood urea nitrogen-to-albumin ratio (BAR) and mortality in critically ill patients with CS. Patients and methods: The data of this retrospective study were extracted from the Medical Information Mart for Intensive CARE-IV (MIMIC-IV v2.2) database. Patients who were diagnosed with CS were included as the study population. The primary outcome was one-year mortality, and secondary outcomes included in-hospital mortality and ICU mortality. Cox regression models were used to estimate the association between BAR and mortality.The prognostic value of BAR was assessed using the time-dependent area under the curve (AUC), net reclassification improvement (NRI), and integrated discrimination improvement (IDI). Results: A total of 1,177 patients were enrolled in the study. After adjustment for covariates, BAR was associated with one-year mortality (HR: 1.02 [1.01–1.03], P < 0.001 per 1-unit increase). Compared with the low-BAR group, the medium- and high-BAR groups showed higher risks of one-year mortality (medium group: HR 1.30 [1.02–1.66], P = 0.034; high group: HR 1.75 [1.32–2.33], P < 0.001). Similar associations were observed for in-hospital mortality (per 1-unit increase: HR 1.02 [1.01–1.04], P < 0.001; high vs. low group: HR 1.51 [1.08–2.13], P = 0.017) and ICU mortality (per 1-unit increase: HR 1.02 [1.01–1.04], P < 0.001; high vs. low group: HR 1.67 [1.17–2.40], P = 0.005). E‑value analysis suggested robustness to unmeasured confounding. BAR demonstrated favorable discriminative ability, with a time‑dependent AUC of 0.612 and a Harrell’s C‑index of 0.607. When combined with existing prognostic indicators, BAR significantly improved prognostic accuracy. Conclusion: In this study, higher BAR levels were associated with increased mortality in CS patients. BAR provided incremental prognostic value when combined with existing indicators. 1. Introduction Cardiogenic shock (CS) is characterized by critically low cardiac output, leadings to severe end-organ hypoperfusion and hypoxia. This acute cardiovascular emergency often presents as multi-organ dysfunction syndrome (MODS), typically compounded by systemic inflammation with profound cellular and metabolic disturbances [ 1 – 3 ]. The primary etiological triad of CS comprises acute myocardial infarction, myocarditis, and acute decompensation of heart failure. Although early revascularization has significantly improved clinical outcomes, mortality rates remain substantial, ranging from 27% to 51% [ 2 – 5 ]. Given the poor prognosis associated with CS, identifying reliable prognostic indicators remains a critical clinical priority. Albumin supports the body in multiple ways, from regulating fluid balance and fighting oxidative stress to easing inflammation. This makes it both a reliable indicator of nutritional status and a key player in maintaining overall organ health [ 6 – 7 ].. Similarly, blood urea nitrogen (BUN) serves as an important indicator, as its elevation is closely tied to critical states like renal hypoperfusion, low cardiac output, dehydration, and neurohormonal activation [ 8 – 9 ].In recent studies, the blood urea nitrogen-to-albumin ratio (BAR) has been reported as a predictor of mortality in intensive care unit (ICU) patients [ 10 – 13 ]. However, its prognostic value in patients with CS remains unevaluated. The high mortality rate of CS underscores the critical need for prognostic indicators. This study therefore aimed to explore the association between BAR and mortality in patients with CS. 2. Methods 2.1. Database The data for this retrospective study were obtained from the Medical Information Mart for Intensive Care-IV (MIMIC-IV v2.2) database. The establishment of this database was approved by the Institutional Review Boards of the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center. Accordingly, the secondary analysis of its de-identified data in this study did not require separate ethical approval. The database includes information on patients admitted to ICUs at Beth Israel Deaconess Medical Center from 2008 to 2019. It contains comprehensive clinical data such as vital signs, laboratory measurements and medications [ 14 – 16 ]. One author (Xu Zhu) obtained access to the database and was responsible for the data extraction (certification number: 41711250). 2.2. Selection of participants This study included all adult patients (≥18 years) diagnosed with CS during hospitalization, based on the International Classification of Diseases, Ninth or Tenth Revision (ICD-9/10) codes. Patients were excluded if they met any of the following criteria: (1) repeated ICU admission during the same hospitalization, (2) length of ICU stay less than 1 day, (3) missing data for BAR or mortality outcomes, or (4) missing data for all covariates included in the multivariable analysis. 2.3. Data extraction Data extraction was performed using PostgreSQL (version 9.6). The variables extracted were demographic data (admission age, sex, ethnicity, admission weight), comorbidities (myocardial infarction, congestive heart failure, hypertension, diabetes, chronic kidney disease, sepsis), laboratory measurements (BUN, albumin, creatinine, lactate, eGFR, calcium, chloride, sodium, potassium, anion gap, bicarbonate, platelet, white blood cell, red blood cell, hemoglobin, glucose), disease severity scores (sequential organ failure assessment score, SOFA score), and treatments (fluid input, mechanical ventilation, renal replacement therapy, antibiotic drugs, vasopressors). All laboratory measurements were defined as the first values recorded within 24 hours of ICU admission. The variable fluid input represented the total volume of infusion within the first 24 hours of ICU admission. Vasopressors included dobutamine, dopamine, epinephrine, norepinephrine, phenylephrine, vasopressin, and milrinone. We also collected hospital and ICU admission/discharge times as well as the date of death. For any variable measured multiple times, the first available value after ICU admission was used in the analysis.. The primary outcome was one-year mortality after ICU admission. Secondary outcomes were in-hospital mortality and ICU mortality. The time origin for all outcomes was defined as the time of ICU admission. For the one-year mortality analysis, patients were censored at one year after ICU admission if alive. For the ICU mortality analysis, patients were censored at ICU discharge if alive. For the in-hospital mortality analysis, censoring occurred at hospital discharge if alive. The Blood Urea Nitrogen to Albumin Ratio (BAR) was calculated as BUN (mg/dL) divided by albumin (g/dL). 2.4. Statistical analysis Study participants were divided into three groups based on BAR tertiles. Continuous variables were described as mean ± standard deviation or (interquartile range, IQR) and compared using one-way ANOVA or the Kruskal-Wallis test, as appropriate. Categorical variables were presented as number (percentage) and compared using the chi-square test. Kaplan-Meier curves with the log-rank test were used to compare one-year mortality across BAR tertile groups. Cox proportional hazards regression was used to quantify hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between BAR and mortality. The proportional hazards assumption was tested using Schoenfeld residuals. Given that the main objective was to estimate the average association over the entire follow-up period, standard Cox models were used when the assumption was satisfied; when it was violated, Cox models with robust standard errors were employed to ensure valid inference [ 17 ]. We constructed two adjusted models: Model I adjusted for age, sex, and ethnicity; Model II additionally adjusted for weight, comorbidities, laboratory measurements, disease severity scores, and treatments. The complete list of variables adjusted for in Model II is provided in the Data Extraction section. The discriminative and incremental prognostic value of BAR was evaluated using the time-dependent area under the receiver operating characteristic curve (AUC), integrated discrimination improvement (IDI), and net reclassification improvement (NRI). Subgroup analyses were performed using stratified Cox regression by admission age, sex, comorbidities, and SOFA score. Interaction terms were used to explore heterogeneity across subgroups. The potential for unmeasured confounding was assessed by calculating the E-value [ 18 ]. A two-tailed P-value < 0.05 was considered statistically significant. All analyses were performed using R software (version 4.3.0) and Empower(R). 3. Results 3.1. Baseline characteristics A total of 1,177 patients were enrolled in the study ( Fig 1 ). The mean age was 69.77 ± 14.63 years, 40.70% of the patients were female and 63.13% were White. Baseline characteristics stratified by BAR tertile are presented in Table 1 . Significant differences across BAR tertiles were observed in age, ethnicity, congestive heart failure, hypertension, diabetes, chronic kidney disease, sepsis, creatinine, lactate, eGFR, chloride, sodium, potassium, anion gap, bicarbonate, platelets, red blood cell count, hemoglobin, glucose, SOFA score, fluid input, use of mechanical ventilation, and renal replacement therapy (all P < 0.05). 10.1371/journal.pone.0352659.t001 Table 1 Baseline characteristics of CS patients. BAR group All patients (0.69 - 58.75) Low (0.69-7.40) Medium (7.41,13.64) High (13.67,58.75) P value Participants, No. 1177 392 390 395 Age (years) 69.77 ± 14.63 65.82 ± 15.93 70.24 ± 14.12 73.22 ± 12.73 <0.001 Sex [female, n (%)] 479 (40.70%) 175 (44.64%) 155 (39.74%) 149 (37.72%) 0.127 Ethnicity[n (%)] 0.011 White 743 (63.13%) 223 (56.89%) 265 (67.95%) 255 (64.56%) Black 111 (9.43%) 37 (9.44%) 34 (8.72%) 40 (10.13%) Other 323 (27.44%) 132 (33.67%) 91 (23.33%) 100 (25.32%) Weight 82.62 ± 22.46 81.87 ± 22.33 83.88 ± 22.73 82.14 ± 22.33 0.396 Comorbidities [n (%)] Myocardial infarction 569 (48.34%) 208 (53.06%) 181 (46.41%) 180 (45.57%) 0.071 Congestive heart failure 919 (78.08%) 282 (71.94%) 305 (78.21%) 332 (84.05%) <0.001 Hypertension 308 (26.17%) 126 (32.14%) 112 (28.72%) 70 (17.72%) <0.001 Diabetes 444 (37.72%) 101 (25.77%) 147 (37.69%) 196 (49.62%) <0.001 Chronic kidney disease 487 (41.38%) 85 (21.68%) 156 (40.00%) 246 (62.28%) <0.001 Sepsis 355 (30.16%) 75 (19.13%) 129 (33.08%) 151 (38.23%) <0.001 Laboratory measurements Creatinine (mg/dL) 1.50 (1.10-2.30) 1.00 (0.80-1.30) 1.50 (1.20-1.90) 2.50 (1.90-3.60) <0.001 Lactate (mmol/L) 2.40 (1.50-4.00) 2.15 (1.40-3.80) 2.30 (1.50-4.00) 2.50 (1.70-4.15) 0.012 eGFR (mL/(min*1.73m 2 )) 42.02 (24.11-63.82) 67.83 (49.94-86.44) 43.22 (31.56-56.20) 21.93 (14.05-30.87) <0.001 Calcium (mg/dl) 8.41 ± 1.04 8.33 ± 1.06 8.50 ± 0.98 8.40 ± 1.09 0.075 Chloride (mmol/L) 100.97 ± 7.48 103.14 ± 7.05 100.83 ± 6.79 98.96 ± 7.96 <0.001 Sodium (mmol/L) 136.98 ± 5.97 137.93 ± 5.78 136.99 ± 5.24 136.02 ± 6.66 <0.001 Potassium (mmol/L) 4.62 ± 1.02 4.26 ± 0.82 4.63 ± 0.99 4.96 ± 1.11 <0.001 Anion gap (mmol/L) 18.51 ± 5.68 16.85 ± 5.02 18.30 ± 5.56 20.36 ± 5.89 <0.001 Bicarbonate (mg/dL) 20.55 ± 5.41 20.69 ± 4.36 21.09 ± 5.51 19.89 ± 6.15 0.006 Platelets (10 9 /L) 205.00 (146.00-272.00) 214.00 (158.75-278.00) 201.50 (148.00-268.00) 190.00 (132.50-264.00) 0.005 White blood cell (10 9 /L) 12.50 (8.70-17.40) 13.00 (8.78-18.35) 12.25 (8.72-17.40) 12.10 (8.70-16.00) 0.115 Red blood cell (10 12 /L) 3.82 ± 0.88 4.06 ± 0.88 3.83 ± 0.84 3.56 ± 0.83 <0.001 Hemoglobin (g/dL) 11.26 ± 2.60 12.10 ± 2.57 11.37 ± 2.51 10.32 ± 2.40 <0.001 Glucose (mg/dL) 154.00 (116.00-219.00) 159.50 (123.00-234.25) 152.00 (116.00-202.50) 147.00 (110.00-218.50) 0.033 Disease severity scores SOFA score 3.00 (1.00-5.00) 2.00 (0.00-4.25) 3.00 (1.00-5.00) 4.00 (1.00-6.00) <0.001 Treatments [n (%)] Fluid input(L) 3.62(1.95-7.14) 4.02 (2.17-8.04) 3.91 (1.96-7.27) 3.07 (1.67-6.22) <0.001 Mechanical ventilation[n (%)] 600 (50.98%) 231 (58.93%) 204 (52.31%) 165 (41.77%) <0.001 Renal replacement therapy[n (%)] 283 (24.04%) 54 (13.78%) 98 (25.13%) 131 (33.16%) <0.001 Antibiotic drugs[n (%)] 1017 (86.41%) 328 (83.67%) 348 (89.23%) 341 (86.33%) 0.076 Vasopressors[n (%)] 1067 (90.65%) 347 (88.52%) 356 (91.28%) 364 (92.15%) 0.189 All laboratory data were the first measurement values taken within 24 hours of admission to the ICU. 10.1371/journal.pone.0352659.g001 Fig 1 Flow chart. 3.2. The association between BAR and outcomes We employed Cox proportional-hazards regression models to assess the associations between BAR and one-year, in-hospital, and ICU mortality. The proportional hazards assumption was met in the in-hospital and ICU mortality models ( P > 0.05 in fully adjusted models) but was violated in the one-year mortality model ( P < 0.05). Therefore, hazard ratios for in-hospital and ICU mortality were obtained from standard Cox models, whereas estimates for one-year mortality were derived using a Cox model with robust standard errors. Results from the time-dependent Cox regression model are included as a sensitivity analysis (Table S1 in S1 File ). After full adjustment, each 1-unit increase in BAR remained associated with one-year mortality (HR: 1.02 [1.01–1.03], P < 0.001), in-hospital mortality (HR: 1.02 [1.01–1.04], P < 0.001), and ICU mortality (HR: 1.02 [1.01–1.04], P < 0.001). When analyzed by BAR tertiles and compared with the low-BAR group, higher BAR levels were associated with increased risk of one-year mortality (medium tertile: 1.30 [1.02–1.66], P = 0.034; high tertile: 1.75 [1.32–2.33], P < 0.001), in-hospital mortality (medium tertile: 1.16 [0.87–1.56], P = 0.311; high tertile: 1.51 [1.08–2.13], P = 0.017), and ICU mortality (medium tertile: 1.30 [0.96–1.77], P = 0.087; high tertile: 1.67 [1.17–2.40], P = 0.005) ( Table 2 ). 10.1371/journal.pone.0352659.t002 Table 2 Univariate and multivariate results by cox regression. Exposure Non-adjusted HR (95%CI) P Adjust I HR (95%CI) P Adjust II HR (95%CI) P One-year mortality BAR per-1unit 1.04 (1.03, 1.04) <0.001 1.03 (1.03, 1.04) <0.001 1.02 (1.01, 1.03) <0.001 BAR group Low 1.0 1.0 1.0 Medium 1.64 (1.33, 2.02) <0.001 1.59 (1.29, 1.97) <0.001 1.30 (1.02, 1.66) 0.034 High 2.62 (2.15, 3.19) <0.001 2.41 (1.97, 2.94) <0.001 1.75 (1.32, 2.33) <0.001 P for trend 1.62 (1.47, 1.78) <0.001 1.55 (1.40, 1.71) <0.001 1.33 (1.15, 1.53) <0.001 In-hospital mortality BAR per-1unit 1.03 (1.02, 1.04) <0.001 1.02 (1.01, 1.03) <0.001 1.02 (1.01, 1.04) <0.001 BAR group Low 1.0 1.0 1.0 Medium 1.28 (0.99, 1.66) 0.059 1.31 (1.01, 1.70) 0.042 1.16 (0.87, 1.56) 0.311 High 1.87 (1.46, 2.38) <0.001 1.78 (1.39, 2.28) <0.001 1.51 (1.08, 2.13) 0.017 P for trend 1.37 (1.22, 1.55) <0.001 1.34 (1.18, 1.51) <0.001 1.24 (1.04, 1.47) 0.014 ICU mortality BAR per-1unit 1.03 (1.02, 1.04) <0.001 1.02 (1.01, 1.03) <0.001 1.02 (1.01, 1.04) <0.001 BAR group Low 1.0 1.0 1.0 Medium 1.31 (1.00, 1.72) 0.046 1.36 (1.04, 1.79) 0.026 1.30 (0.96, 1.77) 0.087 High 1.88 (1.46, 2.43) <0.001 1.80 (1.38, 2.33) <0.001 1.67 (1.17, 2.40) 0.005 P for trend 1.38 (1.21, 1.57) <0.001 1.34 (1.18, 1.52) <0.001 1.29 (1.08, 1.55) 0.005 Non-adjusted model adjusted for none. Adjust I model adjusted for ethnicity, admissionage, sex. Adjust II model adjusted for model I plus weight, myocardial infarct, congestive heart failure, hypertension, diabetes, chronic kidney disease, sepsis, creatinine, lactate, eGFR, calcium, chloride, sodium, potassium, anion gap, bicarbonate, platelets, white blood cell, red blood cell, hemoglobin, glucose,SOFA score, fluid input, mechanical ventilation, renal replacement therapy, antibiotic drugs, vasopressors. Fig 2 illustrates the dose-response associations between BAR and the three mortality outcomes. A positive dose-response relationship was observed for all endpoints. Kaplan-Meier curves for one-year mortality indicated that the high-BAR group had the lowest survival probability (log-rank P < 0.05) ( Fig 3 ). 10.1371/journal.pone.0352659.g002 Fig 2 Dose-response relationships between BAR and mortality endpoints. 10.1371/journal.pone.0352659.g003 Fig 3 Kaplan-Meier curves for one-year mortality of BAR groups. 3.3. Discriminative and Calibrative Performance of BAR To evaluate the discriminative ability of BAR, we calculated the time-dependent AUC for one-year mortality. BAR demonstrated an AUC of 0.612, which was comparable to NT-proBNP (0.608) and lactate (0.595), and higher than the SOFA score (0.544) and troponin T (0.503) ( Fig 4 ). A consistent pattern was observed for Harrell’s C-index, with BAR (0.607) and lactate (0.605) showing similar concordance, followed by NT-proBNP (0.578), the SOFA score (0.543), and troponin T (0.500). The calibration plot of the multivariable Cox model indicated good agreement between observed and predicted probabilities ( Fig 5 ). Time-dependent ROC curves for BUN, albumin, and BAR are presented in Figure S1 in S1 File . 10.1371/journal.pone.0352659.g004 Fig 4 Time-dependent ROC curves for one-year mortality. 10.1371/journal.pone.0352659.g005 Fig 5 Calibration plot of the multivariable Cox model for one-year mortality. 3.4. Incremental clinical value of BAR To evaluate the incremental prognostic value of BAR for one-year mortality over existing indicators, we calculated the net reclassification improvement (NRI) and integrated discrimination improvement (IDI). As shown in Table 3 , BAR significantly improved prognostic performance when added to the SOFA score, lactate, troponin T, or NT-proBNP, compared to using these indicators alone. 10.1371/journal.pone.0352659.t003 Table 3 Incremental clinical value of BAR for one-year mortality. NRI P for NRI IDI P for IDI SOFA vs BAR+SOFA 0.481 <0.001 0.075 <0.001 Lactate vs BAR+Lactate 0.478 <0.001 0.080 <0.001 CTnT vs BAR + CTnT 0.475 <0.001 0.076 <0.001 NT-proBNP vs BAR + NT-proBNP 0.444 <0.001 0.046 <0.001 Specifically, adding BAR to the SOFA score resulted in an NRI of 0.481 ( P < 0.001) and an IDI of 0.075 ( P < 0.001). Similarly, BAR provided statistically significant improvements when added to lactate (NRI = 0.478, P < 0.001; IDI = 0.080, P < 0.001), troponin T (NRI = 0.475, P < 0.001; IDI = 0.076, P < 0.001), and NT-proBNP (NRI = 0.444, P < 0.001; IDI = 0.046, P < 0.001). 3.5. Sensitivity analysis Subgroup analyses were performed based on admission age, sex, SOFA score, myocardial infarction, congestive heart failure, hypertension, diabetes, and chronic kidney disease using Cox models with robust standard errors. The association between BAR and one-year mortality was not statistically significant in the subgroups of age ≤ 65 years, myocardial infarction, hypertension, and diabetes. No significant interaction was observed in any subgroup ( P for interaction > 0.05), indicating that the association between BAR and mortality was generally consistent across these patient characteristics ( Table 4 ). 10.1371/journal.pone.0352659.t004 Table 4 Subgroup analysis for one-year mortality. No. of events/No. of patients HR (95%CI) P P for interaction Age 0.816 ≤ 65 162/396 1.03 (1.00, 1.05) 0.059 > 65 501/781 1.02 (1.01, 1.03) <0.001 Sex 0.404 Female 280/479 1.02 (1.01, 1.04) 0.003 Male 383/698 1.03 (1.01, 1.04) <0.001 SOFA score 0.472 ≤ 3 343/658 1.02 (1.01, 1.04) 0.003 > 3 320/519 1.03 (1.01, 1.04) <0.001 Myocardial infarct 0.349 No 340/608 1.03 (1.02, 1.05) <0.001 Yes 323/569 1.01 (1.00, 1.03) 0.135 Congestive heart failure 0.335 No 153/258 1.03 (1.00, 1.05) 0.020 Yes 510/919 1.02 (1.01, 1.03) <0.001 Hypertension 0.589 No 511/869 1.03 (1.02, 1.04) <0.001 Yes 152/308 1.00 (0.97, 1.03) 0.967 Diabetes 0.607 No 394/733 1.03 (1.02, 1.04) <0.001 Yes 269/444 1.01 (1.00, 1.03) 0.191 Chronic kidney disease 0.900 No 336/690 1.03 (1.01, 1.05) 0.002 Yes 327/487 1.02 (1.01, 1.03) 0.002 To assess the potential impact of unmeasured confounding, an E-value analysis was conducted. The observed association appeared robust, with an E-value of 2.87 for the point estimate. The E-value for the lower confidence interval bound was 1.95, suggesting that to fully explain away the observed association, an unmeasured confounder would need to be associated with both BAR and one-year mortality by relative risks of at least 1.95. 4. Discussion The Blood Urea Nitrogen to Albumin Ratio (BAR) has been associated with prognosis across various diseases, suggesting its broad utility as a prognostic indicator. For instance, BAR has been reported as a useful prognostic indicator for lung cancer patients in intensive care units [ 12 ], the best parameter for predicting outcomes in ICU patients with non-CKD illness [ 13 ], a powerful predictor of in-hospital mortality in older emergency department patients [ 19 ], and a risk factor for mortality in patients with hospital-acquired pneumonia [ 20 ]. In this study, we evaluated the association between BAR and outcomes in critically ill patients with CS. We found that BAR, both as a categorical and continuous variable, was significantly associated with one-year, in-hospital, and ICU mortality in multivariate regression analysis. In time-dependent ROC analysis, BAR demonstrated favorable and comparable discriminative ability. Furthermore, NRI and IDI analyses showed that BAR enhanced the accuracy of prognostic assessment. These findings suggest that BAR may serve as a supplementary prognostic indicator when SOFA score calculation is not feasible. Alternatively, established indicators such as the SOFA score, lactate, NT-proBNP, and troponin T can be used alongside BAR to improve risk stratification. Previous studies have suggested that BUN levels correlate with short-, intermediate-, and long-term prognosis in patients with cardiovascular diseases, including heart failure [ 21 – 25 ]. In CS, low cardiac output and systemic hypoperfusion can lead to reduced renal blood flow, decreased glomerular filtration rate, and impaired urea excretion, thereby increasing BUN levels [ 26 – 28 ]. Additionally, decreased effective circulating volume due to factors such as sweating or vomiting may reduce urine flow and enhance tubular urea reabsorption, further elevating BUN concentration [ 29 – 31 ]. Albumin plays a vital role in maintaining osmotic pressure, regulating vascular permeability and acid-base balance, and functions as an anti-inflammatory and antioxidant molecule. Studies have shown that hypoalbuminemia is a common indicator of poor outcome in CS [ 32 – 33 ]. This condition is frequently complicated by a systemic inflammatory response, which can induce severe cellular and metabolic dysregulation. This inflammatory state may lower albumin levels by impairing hepatic protein synthesis and increasing capillary leakage [ 34 – 37 ]. Taken together, the pathways described above suggest that the prognostic relevance of BAR in CS may be understood through its reflection of two central pathophysiological processes: renal hypoperfusion and systemic inflammation. A high BAR indicates both an elevated BUN concentration, which is associated with reduced renal perfusion, and a low albumin level, which is linked to heightened inflammation. The clinical implications of these findings should be considered alongside the strengths and limitations of our study. This study has several strengths. First, to our knowledge, it is the first to specifically investigate the association between BAR and mortality risk in critically ill patients with CS. Second, since both BUN and albumin are routinely measured with rapid, low-cost tests, BAR represents a feasible and practical indicator for prognostic assessment in clinical practice. Third, the analysis included 1,177 patients from the high-quality MIMIC database; this substantial sample size enhances the reliability of our findings. Finally, BAR primarily reflects a patient’s physiological reserve and metabolic homeostasis, thereby offering a perspective that complements established tools such as the CardShock score or SCAI staging, which focus more on shock severity and clinical progression. Several limitations should be acknowledged. First, BAR in this study was calculated using only the first measurement after ICU admission. Since dynamic changes in biomarkers often carry greater prognostic value than single values [ 38 – 40 ], this approach may underestimate the true association between BAR and mortality. Second, we were unable to account for medications administered prior to ICU admission. Third, due to a high proportion of missing values (exceeding 10%), some variables were excluded from the multivariable regression analysis. Fourth, as a single-center retrospective study, the findings may be subject to selection bias; prospective studies are needed to validate these results. Fifth, key clinical data were unavailable in the database, including detailed etiology, SCAI shock classification, mechanical circulatory support, nutritional support, and hemodynamic parameters, which therefore could not be adjusted for in our analysis. Our findings support BAR as a clinically accessible prognostic indicator that provides additional prognostic information. However, given the observational design, it should not yet guide therapeutic decisions. These findings require validation in prospective, multicenter studies to define BAR’s precise role in clinical management pathways. 5. Conclusions In this retrospective cohort study, elevated BAR at ICU admission was independently associated with increased mortality in patients with CS. When added to existing prognostic indicators, BAR provided significant incremental prognostic value. While these findings highlight BAR’s potential as a supplementary prognostic indicator, its clinical utility requires prospective validation. Supporting information S1 File This file contains Table S1 (The results of the time-dependent Cox regression model) and Figure S1 (Time-dependent ROC curves of BUN, albumin and BAR). (DOC)