Predictive value of blood urea nitrogen to creatinine ratio and estimated plasma volume status in heart failure

Abstract

Background

The blood urea nitrogen to creatinine ratio (BCR) and estimated plasma volume status (ePVS) may be prognostic markers in heart failure (HF), but their combined efficacy is unclear. This research aims to determine the prognostic utility of BCR and ePVS in critically ill HF patients in the intensive care unit.

Methods

Data from the MIMIC-IV database were analyzed. ePVS was calculated using hemoglobin and hematocrit levels with Strauss-derived Duarte method. The primary outcome was 1-year all-cause mortality (ACM). Receiver operating characteristic (ROC) curves identified cutoff values for BCR and ePVS. To assess the connection between BCR, ePVS, and 1-year ACM, the Kaplan-Meier (KM) method, Cox proportional hazards models, subgroup analysis, and limited cubic spline were employed. Harrell’s C statistic evaluated predictive power.

Results

Among 11,066 participants, optimal thresholds for mortality were BCR > 22.81 and ePVS > 7.16 ml/g. BCR demonstrated a non-linear J-shaped correlation with ACM, while ePVS displayed a linear relationship. Multivariate Cox analysis indicated higher level of BCR was linked to higher 1-year ACM (HR = 1.39, 95% CI: 1.30–1.49, P < 0.001), as was increased level of ePVS (HR = 1.09, 95% CI: 1.02–1.16, P = 0.012). Notably, HF patients with both high BCR and ePVS faced a significantly greater mortality risk than those with lower levels of both markers (HR = 1.54, 95% CI: 1.40–1.69, P < 0.001). Combining BCR and ePVS improved prognostic accuracy.

Conclusions

BCR and ePVS independently predict 1-year ACM in HF patients, with their combined use offering improved prognostic accuracy.

Introduction

Heart failure (HF) is a complicated clinical illness defined by severe decreases in cardiac output and/or increases in intracardiac filling pressures, whether at rest or under physical stress. This syndrome results from either poor ventricular filling or blood ejection, or a combination of the two [1]. HF has emerged as a significant clinical and public health concern around the world [2]. The escalating prevalence of HF exerts considerable strain on global healthcare infrastructures. Simple and quick indicators that reflect hemodynamic status have the potential to guide treatment decisions, predict clinical outcomes, and ultimately enhance the management of heart failure.

Blood Urea Nitrogen (BUN) and creatinine (Cr), both renal function markers, are filtered freely at the glomerulus. Post-filtration, however, their renal processing diverges: Cr undergoes minimal reabsorption in the tubules, whereas BUN is significantly reabsorbed, a process modulated by neurohormonal mechanisms. Elevated neurohormonal activity, commonly observed in HF patients, modifies intrarenal dynamics, thereby increasing BUN reabsorption [3]. Accordingly, the BCR is suggested as an indicative marker of poor outcomes in HF [4,5,6], reflecting neurohormonal activity, altered renal perfusion, and other pathophysiological changes [6].

Volume overload represents another critical element of HF pathophysiology, frequently resulting in both hemodynamic and clinical congestion [7]. Effective evaluation of plasma volume (PV) status is crucial for HF management, yet traditional methods like iodine-labeled albumin or carbon monoxide assessments [8, 9] are impractical for routine application due to their complexity. Alternatively, the ePVS, calculated using formulas such as the Strauss-Duarte or Hakim, which incorporate routine laboratory values, offers a simpler and more feasible approach [10,11,12]. This indicator has demonstrated strong associations with clinical outcomes in HF patients, thereby serving as a valuable prognostic tool [12,13,14,15].

Despite the acknowledged utility of both BCR and ePVS individually, their collective role as prognostic indicators in HF is less understood. This study seeks to address this gap by examining the prognostic significance of combined BCR and ePVS levels in predicting long-term outcomes for critically ill HF patients. By analyzing the integrated effects of BCR and ePVS through data from the MIMIC-IV database, this research aims to enhance risk stratification and management strategies for HF, potentially leading to improved clinical outcomes.

Materials and methods

Data source

The data from version 2.2 of the MIMIC-IV database, which is controlled by Beth Israel Deaconess Medical Center. This database provides substantial, publicly available medical records. One author attended the necessary training and acquired certification (No: 58380649) to access the data.

Participants

Participants were critically sick patients in the intensive care unit diagnosed with heart failure, identified using International Classification of Diseases (ICD)-9 and ICD-10 codes. Exclusion criteria included: (1) age below 18 or above 100 years, (2) any subsequent hospitalization rather than the first, and (3) incomplete data on weight, BUN, Cr, hemoglobin, or hematocrit.

Data extraction

Data retrieval was conducted using PostgreSQL and Structured Query Language (SQL) from the MIMIC-IV database [16]. Variables extracted comprised demographic details (age, gender, weight), clinical scores, treatments, vital signs and baseline laboratory values within the initial 24 h of ICU admission. These included mean blood pressure (MBP), heart rate (HR), oxygen saturation (SpO2), white blood cell count (WBC), hemoglobin, hematocrit, platelet count, sodium, potassium, glucose, calcium, BUN, Cr, prothrombin time (PT), international normalized ratio (INR), partial thromboplastin time (PTT). Recorded comorbidities included diabetes, hypertension, myocardial infarction (MI), stroke and chronic kidney disease (CKD). Clinical scores included Sequential Organ Failure Assessment(SOFA), Acute Physiology Score III(APSIII), Oxford Acute Severity of Illness Score(OASIS), Simplified Acute Physiology Score II(SAPS II) and Charlson Comorbidity Index. Clinical interventions such as mechanical ventilation, vasoactive drugs, intra-aortic balloon pump (IABP), loop diuretics, continuous renal replacement therapy (CRRT), antiplatelet drugs, digoxin, statins, guideline directed medical therapy(GDMT) including β-blockers, mineralocorticoid receptor antagonists (MRA), angiotensin II receptor blockers (ARB), and angiotensin-converting enzyme inhibitors (ACEI), angiotensin receptor-neprilysin inhibitor(ARNI) were also documented. We calculated estimated glomerular filtration rate (eGFR) using the CKD Epidemiology Collaboration(CKD-EPI) equation.

ePVS was estimated using the Strauss-derived Duarte formula, calculated as:

ePVS (mL/g) = 100 × (1-hematocrit) ⁄ hemoglobin in g/dL.

Patient outcome indicators

The endpoint was the 1-year all-cause mortality (ACM) rate post-admission.

Statistical analysis

• Variables exhibiting over 20% missing data were excluded from analysis, while multiple imputation was employed to address the remaining missing data [17]. Data were expressed as mean ± SD, while those not normally distributed were described using quartiles. Comparisons of patient characteristics between the deceased and surviving groups were undertaken using the Wilcoxon rank-sum test, Student’s t-test, or Chi-square test, as appropriate.

• ROC curves were generated to determine the best cutoff values for BCR and ePVS based on the greatest Youden index. The area under the curve (AUC) was used to assess the prediction accuracy of BCR and ePVS for 1-year ACM. Subsequently, patients were categorized into groups: high BCR and low BCR, high ePVS and low ePVS. The Kaplan-Meier (KM) method was utilized to estimate the 1-year survival probabilities for these groups. Cox proportional hazards models were implemented to explore associations between various variables and ACM. Three distinct models were developed to analyze the relationship between the indicators (BCR and ePVS) and 1-year ACM. Model I remained unadjusted, Model II included adjustments for age, gender, weight, and comorbidities, and Model III incorporated further adjustments for SOFA, OASIS, Charlson Comorbidity Index, SAPSII, APSIII, HR, MBP, SpO2, WBC, platelet count, PT, PTT, INR, sodium, potassium, glucose, calcium, eGFR, clinical therapies and BCR when analyzing ePVS, ePVS when analyzing BCR. Variance inflation factor (VIF) was used to identify multicollinearity for the Cox proportional hazards models. Variables with VIF above 10 were excluded. The influence of BCR and ePVS was assessed using both categorical and continuous variables, with restricted cubic splines illustrating changes in the relationship between these indicators and the 1-year mortality rate.

• Harrell’s C statistic was used to quantify the improvement in risk prediction for 1-year ACM when BCR and ePVS were included among other factors. A subgroup analysis contrasted the high-BCR + high-ePVS group to the low-BCR + low-ePVS group to investigate the effects of demographic variables and comorbidities on death rates with adjusted Cox proportional-hazards regression analysis. And, sensitivity analyses were conducted as follows: (1) Using the data set(8 333 participants) with complete data; (2) Using data set(11 434 participants) without excluding patients with missing values of weight, BUN, Cr, hemoglobin, or hematocrit.

• Statistical significance was evaluated using a p-value less than 0.05. All statistical analyses were performed using Statistical Package for Social Sciences (SPSS) version 27 and R version 4.2.3.

Results

Baseline features

Figure 1 depicts the flow chart of this study. 11,066 severely sick HF patients participated in this study. The cohort’s median age was 73.83 ± 13.42 years. The majority of 6,095 (55.08%) were males.

Fig. 1

Selection of study population from MIMIC-IV database

Table 1 compares the baseline characteristics of patients who survived and those who did not. The deceased group was generally older, had lower body weight, and comprised a lower percentage of males. This group also showed higher prevalence of MI, stroke and CKD, and more frequent utilization of CRRT, digoxin and vasoactive drugs. In contrast, they had lower rates of hypertension and were less likely to be on loop diuretics, statin, antiplatelet drugs and GDMT. Significant differences were also noted in the Charlson Comorbidity Index, OASIS, SOFA score, APSIII, SAPSII, HR, WBC, potassium, glucose levels, PT, PTT, INR, BUN, Cr, BCR, and ePVS, which were all higher in the deceased group. Conversely, MBP, SpO2, calcium, eGFR, hemoglobin, and hematocrit levels were lower in the deceased group (all p < 0.05).

Table 1 Demographic and clinical characteristics of patients with heart failure

Association between BCR and ePVS with 1-Year ACM

ROC curves were employed to establish cut-off values for BCR and ePVS as predictors of 1-year ACM (Fig. 2). The AUC for BCR was determined to be 0.59 with an optimal cut-off at 22.81. For ePVS, the AUC was 0.56, with a threshold set at 7.16. Table S1 shows the baseline characteristics of patients with different levels of BCR and ePVS.

Fig. 2

ROC curves to predict 1-year ACM.(A) BCR and 1-year mortality rate; (B) ePVS and 1-year mortality rate. BCR, blood urea nitrogen to creatinine ratio; ePVS, estimated plasma volume status

The restricted cubic spline analysis revealed a nonlinear J-shaped relationship between BCR and 1-year ACM (overall P < 0.001, nonlinearity P < 0.001) (Fig. 3B). The mortality risk was minimized at a BCR of 14.23, below which the associations were inversely correlated and above which they were directly correlated. A linear relationship was observed for ePVS and 1-year ACM (overall P = 0.034, nonlinearity P = 0.202) as presented in Fig. 3D. Elevated values above the identified cut-offs (BCR: 22.81, ePVS: 7.16) corresponded to a heightened risk of mortality, as demonstrated by the cubic spline curve.

Fig. 3

Association between indicators and 1-year all-cause mortality.(A)Unadjusted model for BCR; (B) Adjusted model for BCR; (C)Unadjusted model for ePVS; (D) Adjusted model for ePVS. BCR, blood urea nitrogen to creatinine ratio; ePVS, estimated plasma volume status

Survival analysis

Patients were divided into groups based on their BCR and ePVS levels, which were determined by the predefined cutoff values: high BCR (n = 4,310) and low BCR (n = 6,756), as well as high ePVS (n = 4,436) and low ePVS (n = 6,630). The characteristics of each group was in Table S1. Furthermore, the combination of these measurements divided patients into four groups: low-BCR + low-ePVS (n = 4,165), high-BCR + low-ePVS (n = 2,465), low-BCR + high-ePVS (n = 2,591), and high-BCR + high-ePVS (n = 1,845).

The 1-year ACM rates were evaluated among these groups, demonstrating that the low BCR group suffered considerably decreased mortality(p < 0.001) (Figure S1). A similar pattern was observed for the low ePVS group (p < 0.001) (Figure S2). Figure 4 shows that low-BCR + low-ePVS patients had the lowest death rate, whereas the patients with high-BCR + high-ePVS had the highest, with significant statistical differences (P < 0.001).

Fig. 4

Kaplan-Meier survival curve for 1-year all-cause mortality of different groups. BCR, blood urea nitrogen to creatinine ratio; ePVS, estimated plasma volume status

Cox proportional hazard models were used to investigate the connections between BCR, ePVS, and death. INR was excluded in the adjusted model III with VIF > 10. As shown in Table 2, high BCR was associated with a significantly higher death risk (HR = 1.65, 95% CI: 1.55–1.76, p < 0.001). After adjusting for several confounders in Model III, the high BCR group still had higher death rates (HR = 1.39, 95% CI: 1.30–1.49, p < 0.001). Initially, greater ePVS levels were associated with a substantial increase in mortality (HR = 1.34, 95% CI: 1.26–1.43, p < 0.001), which remained after full correction (HR = 1.09, 95% CI: 1.02–1.16, p = 0.012).

Table 2 Association between the BCR and ePVS with 1-year all-cause mortality

As shown in Table 3, in univariate models, persons with high-BCR + low-ePVS, low-BCR + high-ePVS, and high-BCR + high-ePVS had significantly higher death risks (HRs = 1.74, 1.41, and 2.13 respectively; all p < 0.001). After correction in Model III, the increased risks remained substantial for those with high-BCR + low-ePVS and high-BCR + high-ePVS (HRs = 1.37 and 1.54, respectively; both p < 0.001).

Table 3 Association between the ePVS + BCR and 1-year all-cause mortality

Predictive performance of BCR and ePVS for 1-Year ACM

The predictive efficacy of BCR and ePVS for 1-year ACM is showed in Table S2. The unadjusted model of BCR and ePVS yielded a C-index for all‐cause mortality of 0.5703 and 0.5476, respectively, which rose to 0.5855 when combined together(p < 0.05). The inclusion of BCR enhanced the C-index of a fully adjusted model from 0.7608 to 0.7643 (p < 0.001), while the addition of ePVS raised it to 0.7614 (p = 0.005). The combined incorporation of both indicators yielded the highest C-statistic of 0.7646, demonstrating superior predictive power over the model without these indicators and models including either indicator alone (all p for comparison < 0.05).

Subgroup analysis of patients with high-BCR + high-ePVS versus low-BCR + low-ePVS

In the subgroup analysis depicted in Fig. 5, after adjusting for potential confounders, the interaction was significant among the subgroups of stroke, CKD, and eGFR (all P for interaction < 0.05). Patients with high-BCR + high-ePVS generally exhibited increased mortality across most subgroups when compared with low-BCR + low-ePVS, except in subgroup of patients with stroke, where the differences were not statistically significant (P = 0.516).

Fig. 5

Subgroup analysis between HF patients with high-BCR + high-ePVS and low-BCR + low-ePVS. HF, heart failure; BCR, blood urea nitrogen to creatinine ratio; ePVS, estimated plasma volume status

Sensitivity analysis

Robust outcomes were obtained from the two sensitivity analysis datasets (Table S3–6). Both a higher level of BCR and higher ePVS were significantly associated with an elevated all - cause mortality rate. In comparison to the other three groups, the high-BCR + high-ePVS group demonstrated the highest mortality rate. Among the 8333 patients with complete data, the unadjusted and adjusted model incorporating BCR and ePVS exhibited the highest predictive efficacy, as indicated by the highest C-statistic index (Table S7).

Discussion

This investigation underscores the significant prognostic utility of BCR and ePVS at admission for forecasting 1-year ACM among critically ill HF patients. The main finding of this study was that BCR and ePVS were both significantly correlated with the 1-year ACM. And the all-cause mortality was lowest in the low-BCR + low-ePVS group and highest in the high-BCR + high-ePVS group. Integrating the assessment of both BCR and ePVS proves more effective than evaluating either metric independently in predicting outcomes. This is the inaugural study delineating the pivotal correlations between these markers and the long-term outcomes in this patient demographic. The findings suggest that the combination of BCR and ePVS may be viable tools for enhancing risk stratification and improving management strategies for HF patients.

In contrast to Cr, BUN is reabsorbed through the tubules. In HF patients, decreased cardiac output and inadequate arterial filling activate the sympathetic nervous system (SNS), as well as the renin-angiotensin-aldosterone system (RAAS), encouraging water and sodium retention and enhanced passive urea reabsorption in the tubules [3]. Analysis of data from 427 participants in the DOSE-AHF and CARRESS-HF studies demonstrated that people with higher-than-median baseline RAAS indicators had lower blood pressure and higher BUN levels [18]. Furthermore, clinical studies have indicated that arginine vasopressin (AVP) levels are frequently high in HF patients, which promotes BUN reabsorption in the collecting ducts [19]. Elevated BCR corresponds with poorer outcomes in both acute and chronic HF (AHF and CHF) and serves as an independent predictor of ACM [5, 6, 20, 21]. Our analysis confirms that a BCR cutoff value of 22.81 effectively predicts mortality, as demonstrated through ROC curve analysis. Other studies on BCR and heart failure have primarily used BCR thresholds ranging from 15.32 to 25.5 [22]. Adjusting for confounders via Cox regression, a BCR>22.81 correlates with higher mortality (HR = 1.39, 95% CI: 1.30–1.49, p < 0.001). These results resonate with prior research; for example, a prospective cohort study indicated that a BUN/Scr ratio ≥ 25.09 independently predicts long-term mortality in AHF patients [21]. Similarly, in AHF, Matsue et al. found that a BUN/Cr ratio above the normal range independently predicts poor outcomes, including ACM and cardiovascular or renal rehospitalization [20]. The significance of BCR in chronic heart failure (CHF) has also garnered attention. Wang et al. emphasized the importance of BCR in CHF, demonstrating that a ratio exceeding 19.37 is an independent predictor of mortality [5].

Moreover, we identified a nonlinear J-shaped relationship between BCR and 1-year ACM, pinpointing a BCR of 14.23 as associated with the lowest mortality risk. The PROTECT study noted a median BCR of 15.0 in a general population devoid of cardiovascular risk factors [20]. Shen et al. reported a nonlinear U-shaped relationship between BCR and mortality in the general population [23].Very low BCR levels might indicate tubular damage [24] or reduced protein intake, while extremely high levels could reflect advanced renal dysfunction and severe neurohormonal activation, both of which portend a grim prognosis. Thus, maintaining an optimal BCR range might reflect better management of HF and its related renal complications.

Volume overload and fluid congestion pose substantial obstacles in controlling HF. Kobayashi et al. found that ePVS, derived using Duarte’s method, significantly enhanced predictive performance in patients with HF with preserved ejection fraction (HFpEF) [12]. In contrast, ePVS derived using Hakim’s method did not exhibit similar prognostic significance. In our investigation, we used Duarte’s technique to calculate ePVS, which revealed a linear association between ePVS and ACM. We found a significant connection between higher ePVS and increased mortality rates using Cox proportional hazard models (HR = 1.09, 95% CI: 1.02–1.16, P = 0.012). The predictive importance of ePVS for HF in our investigation is consistent with earlier findings. For example, Balderston et al. established ePVS as a predictor of death or hospital readmission in HF patients, independent of brain natriuretic peptide(BNP) levels [14]. Furthermore, Huang et al. reported that higher baseline ePVS correlated with increased mortality and HF hospitalization rates after a median follow-up of 10.7 years [25]. They have reported that a threshold of higher than 5.5 ml/g of ePVS was associated with fluid overload status and increased mortality in HF patients [25]. In our study, the average ePVS was 6.82 ml/g, with a cutoff value of 7.16 for ACM. These values are higher than those reported in previous research [12, 26, 27], which could be ascribed to our group of critically sick patients in the intensive care unit (ICU) who presented with significant fluid overload and deteriorated physiological condition. A recent study using data from the MIMIC-IV database on patients who underwent coronary revascularization found significant higher ePVS values (mean value 7.38 ± 2.42) [28], which aligns with our findings.

In our study, the AUC values of BCR and ePVS were 0.59 and 0.56, respectively. These values, although not remarkably high, can be attributed to the complex interplay of multiple factors such as age, the severity of the disease, comorbid conditions, and treatment modalities influencing heart failure prognosis. As a result, the predictive capacity of a solitary laboratory parameter, such as BCR or ePVS, is inherently limited. Notwithstanding this limitation, when considering BCR and ePVS, either as continuous variables or as categorical variables stratified by cutoff values, both the Kaplan - Meier (KM) curve analysis and the multivariate Cox regression model consistently demonstrated their substantial predictive value for the risk of mortality.

The patients were categorized into 4 distinct profiles: low-BCR + low-ePVS, low-BCR + high-ePVS, high-BCR + low-ePVS and high-BCR + high-ePVS. Both Kaplan-Meier curves and the adjusted Cox proportional hazards regression model showed a highest 1-year mortality rate in the high-BCR + high-ePVS group. After accounting for potential confounding factors, these patients demonstrated significantly increased mortality risk (HR = 1.54, 95% CI: 1.40–1.69, p < 0.001), consistent across most subgroups analyzed. Further examination into the combined prognostic capability of BCR and ePVS for HF outcomes revealed a substantial enhancement in prognostic accuracy, as indicated by an increased C-index. The high-BCR + high-ePVS status may represent a condition with both hypoperfusion and congestion. By combining ePVS with BCR, clinicians can achieve a multidimensional perspective encompassing both neurohormonal dynamics and volume status, leading to more tailored therapeutic strategies. Moreover, the biomarkers utilized for calculating BCR and ePVS are not only simple, rapid, and cost-effective but also readily integrate into clinical practice, enhancing their practical utility. Similarly, Nogi et al. presented a new classification of 4 hemodynamic profiles using the fractional excretion of urea nitrogen(FEUN) and ePVS values, they found that the low-FEUN/high-ePVS group had a higher mortality than the high-FEUN/low-ePVS in patients with ADHF [29]. In the unadjusted model, the C-indices for BCR and ePVS indicated that their combination yielded a significantly higher C-index compared to each variable analyzed in isolation(BCR:0.5703, ePVS:0.5476, BCR + ePVS:0.5855). While, within the adjusted model, the incremental increase in the C-index was less pronounced. This subdued improvement might be attributed to the intricate interplay of numerous factors that collectively influence the prognosis of heart failure. Each of these factors likely contributes to the overall outcome in a complex manner, thereby dampening the additional predictive power gained from combining BCR and ePVS. Nonetheless, it is important to note that even this seemingly modest enhancement in the prognostic efficacy was statistically significant. This finding suggests that, despite the complexity introduced by multiple prognostic factors, the combination of BCR and ePVS still holds value in refining the prediction of heart failure outcomes.

BUN is not a reliable indicator of eGFR, and the prognostic value of BCR for predicting poorer outcomes in HF is independent of eGFR. In our study, we analyzed differences in renal function between groups with high and low BCR and observed that the high BCR group demonstrated lower creatinine levels and higher eGFR levels(Table S1). Furthermore, the influence of BCR on all-cause mortality (ACM) remains significant even after adjusting for eGFR, aligning with previous research. Additionally, there was a higher incidence of loop diuretic use within the high BCR group, likely due to greater congestion levels in these patients [30], and the use of diuretic might also increase BCR levels. In contrast, patients with higher ePVS had significantly higher creatinine levels, lower eGFR, and a greater prevalence of chronic kidney disease (CKD), which potentially contributed to increased water and sodium retention [31]. Notably, ePVS was independently associated with higher mortality rates in HF patients after adjusting for eGFR. Our subgroup analysis comparing high-BCR + high-ePVS with low-BCR + low-ePVS indicated significant heterogeneity across stroke, CKD, and eGFR subgroups (all P for interaction < 0.05). Patients with both elevated BCR and ePVS exhibited significantly higher one-year ACM across all eGFR subgroups, as well as among patients with or without CKD. Within the stroke subgroup, the mortality rate among patients with high-BCR + high-ePVS was higher than that among those with low-BCR + low-ePVS (52.4% and 39.0%, respectively). This association was attenuated in stroke patients after adjustment, likely attributable to the small sample size of stroke patients(n = 619) and unmeasured confounders. From a mechanistic perspective, the impact of high volume state on cerebral perfusion, neurogenic myocardial depression [32] and therapeutic interventions(e.g. mannitol) in stroke patients may weaken the association between volume indicators and prognosis. The impact of CKD and different levels of eGFR on urea metabolism and volume can also lead to interactions. This heterogeneity is clinically instructive that in complex subgroups (such as stroke and CKD), prognostic assessment requires the integration of neurological function indicators and dynamic volume monitoring to overcome the limitations of traditional biomarkers. It should be noted that these results are exploratory and need to be validated in prospective studies.

Despite these encouraging results, there are several limitations that must be acknowledged. First, the nature of this study being a single-center retrospective observational study restricts its applicability across different demographics and medical settings. To enhance the reliability and applicability of these findings, future investigations should aim at conducting prospective, multi-center randomized controlled trials. Second, the short ICU stay duration(3.95 ± 5.18 days) in our cohort limited analyzing dynamic trends in laboratory indicators or account for time-dependent confounders, which may affect outcome interpretation. Third, our analysis was unable to incorporate all conceivable confounding factors, notably the NT-proBNP levels and left ventricular ejection fraction (LVEF), due to incomplete data availability. This omission may have compromised the comprehensive exclusion of potential confounders, thereby impacting the precision and dependability of our findings. Additionally, the ePVS was calculated using Strauss’ formula. Hakim’s method which requires dry weight could not be applied. Future studies comparing these approaches are warranted to evaluate their differential prognostic implications. Future research addressing these gaps will be essential to solidify the prognostic utility of both BCR and ePVS for clinical application.

Conclusion

To summarize, this retrospective examination of an extensive database confirms that both BCR and ePVS independently predict long-term mortality in patients with HF. Utilizing these markers in conjunction can improve prognostic accuracy beyond the application of each biomarker singly. These biomarkers, which are both readily accessible and cost-effective, markedly enhance our capacity for risk stratification and the customization of treatment approaches, thus potentially improving patient outcomes in a clinical context.