Abstract
Background
Sepsis has posed a significant global public health challenge, with advanced age, chronic diseases, and medication history in the older adults population significantly increasing the risk of organ failure or death. Serum creatinine (Cr) and albumin (Alb) are important predictors of mortality in individuals with various diseases. Therefore, our research aimed to evaluate the relationship between serum creatinine to albumin ratio (CAR) and 28-day all-cause mortality in older adults patients with sepsis.
Methods
Using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, we identified older adults patients with sepsis and applied Kaplan-Meier (K-M) curves and multivariable logistic regression to evaluate the association between CAR and 28-day all-cause mortality. Restricted cubic spline (RCS) analysis explored the nonlinear relationship, and subgroup analyses were conducted.
Results
After applying inclusion and exclusion criteria, a total of 2,350 older adults patients with sepsis were enrolled, among whom the proportion of non-survivors was 34.89%. Survivors had longer hospital stays and lower CAR. Patients in the highest CAR quartile (Q4) exhibited significantly higher levels of creatinine, and various disease scores. Additionally, the 28-day all-cause mortality in the ICU for the Q4 group was significantly higher than in the other groups (p < 0.0001). After adjusting for confounding factors, the 28-day ICU all-cause mortality was still significantly increased, with a nonlinear relationship (P < 0.001). This association was further confirmed in subgroup analyses, where the predictive value of CAR for 28-day all-cause mortality in the ICU remained consistent across subgroups stratified by comorbidities and individual patient characteristics.
Conclusions
The MIMIC-IV database revealed a positive association between serum CAR and short-term all-cause mortality in older adults patients with sepsis, with a nonlinear relationship. This research enhances our understanding of the association between serum-based biomarkers and prognosis in older adults patients with sepsis.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12877-025-06093-3.
Keywords: MIMIC-IV, Serum creatinine to albumin ratio, Sepsis, All-cause mortality
Background
Sepsis was a life-threatening clinical syndrome characterized by a dysregulated host response to infection leading to organ dysfunction, posing a significant global healthcare burden. It was estimated that the annual incidence of sepsis exceeded 50 million cases worldwide, with mortality rates ranging from 20 to 50% depending on disease severity [1]. Sepsis imposed substantial economic and societal costs, particularly in aging populations. Older adults were more severely impacted by sepsis due to age-related immunosenescence, increased prevalence of comorbidities such as diabetes, cardiovascular diseases, and chronic kidney disease, combined with alterations in inflammatory response regulation [2]. These interacting factors exacerbated pathophysiological cascades, leading to enhanced systemic inflammation, impaired immune resolution, and increased risk of multiple organ failure [3]. Given that the global population aged 65 years and older was projected to double by 2050, the clinical and public health significance of sepsis management became increasingly prominent, underscoring the growing urgency for effective interventions.
Despite advancements in antimicrobial therapy and supportive care within intensive care units, sepsis mortality remained high, primarily due to delayed identification of high-risk patients and inadequate risk stratification tools. Previous prognostic methods depended on clinical scoring systems (e.g., Sequential Organ Failure Assessment (SOFA), qSOFA) and individual biomarkers (e.g., procalcitonin, lactate), which had substantial limitations in predicting outcomes among complex patient populations. There was an urgent need to identify reliable and accessible biomarkers reflecting multipathological dimensions, encompassing organ damage, inflammatory burden, nutritional status, and other factors, driving the advancement of research on composite indices [4].
The serum creatinine-to-albumin ratio (CAR) was a promising candidate due to its physiological relevance to sepsis pathophysiology. Serum creatinine, a common marker of renal dysfunction, directly reflected sepsis-associated acute kidney injury (AKI), which occurred in 30–50% of severe sepsis cases and independently predicted poor outcomes [5, 6]. Additionally, albumin served as a surrogate for nutritional status and inflammatory activity; hypoalbuminemia in sepsis, caused by increased vascular permeability, hepatic synthetic dysfunction, and catabolic states, was associated with prolonged illness and mortality [7, 8]. By integrating these two parameters, CAR theoretically captured the dual pathophysiological axes, renal injury and inflammation-nutrition dysregulation, central to sepsis progression. Findings from several clinical studies demonstrated associations between CAR and disease severity in critical illnesses such as acute respiratory distress syndrome and traumatic brain injury [9–11]. However, its role in sepsis-specific prognosis remained underexplored, particularly in older populations characterized by complex comorbidities [12, 13].
Given the urgent need for reliable prognostic tools and the potential value of the creatinine-to-albumin ratio (CAR) in reflecting sepsis-related pathobiology, the MIMIC-IV database provided a unique opportunity to explore the association between CAR and clinical outcomes in a critically ill patient cohort. This study aimed to evaluate the relationship between CAR and 28-day all-cause mortality in older adults with sepsis. Such an investigation would help elucidate underlying pathophysiological mechanisms, facilitate early risk stratification, and potentially inform the development of novel risk stratification strategies. The research held important implications for clinical practice, potentially guiding resource allocation and facilitating the implementation of targeted interventions to improve outcomes in this high-risk population.
Materials and methods
Study design overview
Our research was a retrospective cohort study based on the MIMIC-IV database, aiming to investigate the association between the CAR and 28-day all-cause mortality in older adults with sepsis, and data were extracted from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. MIMIC-IV database is a publicly available resource containing electronic health records from intensive care unit (ICU) patients. It includes medical data from the Beth Israel Deaconess Medical Center (BIDMC) spanning from 2008 to 2019. Approved by the Institutional Review Board (IRB), the database is exempt from requiring additional informed consent or ethical clearance [14].
To gain access to the MIMIC-IV database, we completed the required online courses and exams (record ID: 64623062). From the database, we identified 5,303 sepsis patients aged 65 to 90 years. After excluding patients with ICU stays shorter than 24 h (n = 1,354), patients with missing data (n = 1,557), and those with a SOFA score below 2 (n = 42), 2,350 patients remained. These patients were then stratified based on 28-day clinical outcomes and serum CAR quartiles (Q1 = 572, CAR ≤ 0.33; Q2 = 542, 0.33 < CAR ≤ 0.50; Q3 = 648, 0.50 < CAR ≤ 0.89; Q4 = 588, CAR > 0.89). A detailed data extraction process is shown in the flowchart (Fig. 1).
Fig. 1.
Flowchart for participants from the MIMIC-IV
Data source and collection
MIMIC-IV Database was acquired with proper authorization, the demographic characteristics included age, sex, body mass index (BMI), and race. Vital signs included heart rate (HR), respiratory rate (RR), mean arterial pressure (MAP), systolic blood pressure (SBP), diastolic blood pressure (DBP), and body temperature (BT). Laboratory test results included white blood cell count (WBC), platelet count, hemoglobin, creatinine, albumin, blood urea nitrogen (BUN), total bilirubin (TB), potassium (K), sodium (Na), calcium (Ca), and anion gap. Disease scores included Simplified Acute Physiology Score II (SAPS II), and Sequential Organ Failure Assessment (SOFA). Documented comorbidities included acute kidney injury (AKI), respiratory failure (RF), heart failure (HF), atrial fibrillation (AF), hepatopathy, and nephropathy. Medications included vasopressin, octreotide acetate, and statins such as atorvastatin, aspirin, and heparin. Treatment included dialysis presence, non-invasive mechanical ventilation, and extracorporeal membrane oxygenation (ECMO). Clinical outcomes included ICU length of stay (LOS ICU), hospital length of stay (LOS hospital), and 28-day all-cause mortality in the ICU. The initial recorded values were determined when the variables were documented more than once in the previous 24 h [15].
Statistical analysis
In the MIMIC-IV cohort, CAR was computed as both a continuous and a categorical variable (quartiles: Q1-Q4), and incorporated into the logistic regression model as a categorical variable. Binary logistic regression modeling was used to assess the association of Q2-Q4 with mortality, with Q1 as the reference, to evaluate its association with 28-day all-cause mortality in older adults patients with sepsis. Continuous variables were expressed as mean ± standard deviation (SD) for normally distributed variables, and numbers (%) for categorical variables. Analysis of variance (ANOVA) and χ2 tests were used for group comparisons [16].Kaplan-Meier (K-M) survival curves and log-rank tests were used to analyze differences in 28-day all-cause mortality among the four groups [17].Multivariable logistic regression analysis was conducted to determine the association between serum CAR and 28-day all-cause mortality in older adults patients with sepsis. Two models adjusted for covariates were assessed for the cohort (Model 1, unadjusted; Model 2, adjusted for age, BMI, WBC, platelet count, hemoglobin, BUN, TB, Na, Ca, K, anion gap, MAP, HR, RR, SBP, DBP, BT, SAPS II, SOFA, aspirin, atorvastatin, heparin, octreotide acetate, vasopressin, non-invasive mechanical ventilation, HF, RF, hepatopathy, nephropathy, and dialysis presence) [18, 19].
Restricted cubic spline (RCS) analysis was employed to explore the nonlinear relationship between serum CAR and clinical outcomes in older adults patients with sepsis [20].
The subgroup analyses focused on specific populations. In this research attention was given to age (< 80 and ≥ 80 years), sex (female and male), race (black, white, and other), BMI(< 30 kg/m² and ≥ 30 kg/m²), WBC, platelet count, hemoglobin, BUN, TB, Na, Ca, K, anion gap, MBP, SBP, DBP, HR, RR, BT, SAPSII, SOFA, aspirin, atorvastatin, heparin, octreotide acetate, vasopressin, mechanical ventilation, heart failure, respiratory failure, hepatopathy, nephropathy, dialysis present [11, 21].
Results were expressed as odds ratios (OR) and 95% confidence interval (CI). Statistical analysis was performed using R software (version 4.4.1). p < 0.05 was considered statistically significant.
Results
Baseline characteristics
A total of 2,350 patients were included based on the application of inclusion and exclusion criteria. They were divided into quartiles according to their admission serum CAR levels (Q1 = 572, CAR ≤ 0.33; Q2 = 542, 0.33 < CAR ≤ 0.50; Q3 = 648, 0.50 < CAR ≤ 0.89; Q4 = 588, CAR > 0.89) (Table 1). Significant differences across CAR quartiles were observed for 28-day all-cause mortality, race, gender, age, ICU length of stay, BMI, platelet count, hemoglobin, TB, Na, Ca, BT, and aspirin use (p < 0.05). From Q1 to Q4, there was a gradual increase in 28-day all-cause mortality, height, WBC count, creatinine, BUN, potassium, anion gap, SAPS II, and SOFA scores, while albumin, MAP, and DBP showed a gradual decrease. In the Q4 CAR group, a higher proportion of patients received vasopressin and dialysis, while fewer used octreotide acetate. Additionally, patients in Q4 had a higher prevalence of hepatopathy and nephropathy (p < 0.05).
Table 1.
Baseline characteristics of participants stratified by CAR quartiles
| Level | Q1 | Q2 | Q3 | Q4 | p | |
|---|---|---|---|---|---|---|
| n | 572 | 542 | 648 | 588 | ||
| Death (%) | No | 440 ( 76.9) | 372 ( 68.6) | 395 ( 61.0) | 299 ( 50.9) | < 0.001 |
| Yes | 132 ( 23.1) | 170 ( 31.4) | 253 ( 39.0) | 289 ( 49.1) | ||
| race (%) | BLACK | 43 (7.5) | 54 ( 10.0) | 67 ( 10.3) | 87 ( 14.8) | 0.001 |
| OTHER | 141 ( 24.7) | 129 ( 23.8) | 131 ( 20.2) | 144 ( 24.5) | ||
| WHITE | 388 ( 67.8) | 359 ( 66.2) | 450 ( 69.4) | 357 ( 60.7) | ||
| gender (%) | F | 302 ( 52.8) | 236 ( 43.5) | 274 ( 42.3) | 232 ( 39.5) | < 0.001 |
| M | 270 ( 47.2) | 306 ( 56.5) | 374 ( 57.7) | 356 ( 60.5) | ||
| age (mean (SD)) | 75.46 (7.06) | 77.07 (6.82) | 76.88 (6.73) | 75.85 (6.77) | < 0.001 | |
| Los icu (mean (SD)) | 6.64 (7.63) | 6.30 (6.46) | 6.27 (6.60) | 7.48 (9.86) | 0.025 | |
| Los hospital (mean (SD)) | 15.02 (14.24) | 14.73 (12.96) | 14.80 (15.97) | 15.54 (17.23) | 0.797 | |
| weight (mean (SD)) | 73.88 (21.07) | 81.64 (22.06) | 83.95 (22.79) | 83.17 (22.93) | < 0.001 | |
| height (mean (SD)) | 166.47 (9.51) | 167.80 (9.73) | 168.28 (9.44) | 169.01 (9.21) | < 0.001 | |
| BMI (mean (SD)) | 26.65 (7.35) | 29.06 (7.84) | 29.72 (8.11) | 29.19 (8.17) | < 0.001 | |
| WBC (mean (SD)) | 11.79 (8.50) | 12.30 (7.93) | 13.24 (12.67) | 13.68 (9.49) | 0.005 | |
| platelets (mean (SD)) | 193.86 (113.33) | 174.34 (105.00) | 174.28 (110.54) | 169.35 (109.29) | 0.001 | |
| hemoglobin (mean (SD)) | 9.66 (2.04) | 9.71 (2.22) | 9.33 (2.00) | 8.93 (1.87) | < 0.001 | |
| creatinine (mean (SD)) | 0.74 (0.22) | 1.17 (0.28) | 1.76 (0.50) | 3.86 (1.81) | < 0.001 | |
| albumin (mean (SD)) | 3.04 (0.64) | 2.91 (0.60) | 2.70 (0.59) | 2.54 (0.64) | < 0.001 | |
| BUN (mean (SD)) | 19.01 (10.10) | 29.55 (14.53) | 42.31 (21.33) | 63.59 (33.13) | < 0.001 | |
| Bilirubin total (mean (SD)) | 1.18 (2.05) | 1.40 (1.80) | 1.76 (3.01) | 1.67 (4.02) | 0.002 | |
| sodium (mean (SD)) | 136.34 (5.37) | 136.23 (5.82) | 136.92 (6.61) | 135.29 (6.68) | < 0.001 | |
| calcium (mean (SD)) | 7.72 (0.86) | 7.80 (0.82) | 7.72 (0.85) | 7.51 (1.08) | < 0.001 | |
| potassium (mean (SD)) | 3.64 (0.54) | 3.89 (0.61) | 3.96 (0.64) | 4.17 (0.71) | < 0.001 | |
| Anion gap (mean (SD)) | 11.94 (3.00) | 13.14 (3.61) | 14.28 (4.08) | 16.80 (4.44) | < 0.001 | |
| CAR (mean (SD)) | 0.24 (0.06) | 0.40 (0.05) | 0.65 (0.11) | 1.57 (0.79) | < 0.001 | |
| MBP (mean (SD)) | 74.32 (8.30) | 73.35 (8.01) | 72.06 (7.48) | 71.39 (8.29) | < 0.001 | |
| Heart rate (mean (SD)) | 89.60 (16.67) | 88.95 (17.41) | 90.62 (18.47) | 89.78 (17.91) | 0.436 | |
| Resp rate (mean (SD)) | 20.86 (3.92) | 21.24 (3.98) | 21.49 (3.94) | 21.18 (4.55) | 0.068 | |
| SBP (mean (SD)) | 109.69 (12.11) | 109.53 (11.51) | 108.36 (11.52) | 108.36 (11.43) | 0.079 | |
| DBP (mean (SD)) | 60.33 (8.74) | 59.07 (8.83) | 57.51 (8.22) | 56.85 (8.81) | < 0.001 | |
| temperature (mean (SD)) | 36.96 (0.49) | 36.93 (0.59) | 36.87 (0.58) | 36.73 (0.63) | < 0.001 | |
| SAPSII (mean (SD)) | 45.20 (13.12) | 48.48 (12.73) | 53.58 (14.19) | 57.81 (14.02) | < 0.001 | |
| SOFA (mean (SD)) | 6.69 (3.00) | 7.49 (3.26) | 8.78 (3.57) | 10.40 (3.52) | < 0.001 | |
| Aspirin (%) | No | 359 ( 62.8) | 306 ( 56.5) | 356 ( 54.9) | 325 ( 55.3) | 0.022 |
| Yes | 213 ( 37.2) | 236 ( 43.5) | 292 ( 45.1) | 263 ( 44.7) | ||
| Atorvastatin (%) | No | 421 ( 73.6) | 367 ( 67.7) | 451 ( 69.6) | 405 ( 68.9) | 0.152 |
| Yes | 151 ( 26.4) | 175 ( 32.3) | 197 ( 30.4) | 183 ( 31.1) | ||
| Heparin (%) | No | 83 ( 14.5) | 62 ( 11.4) | 92 ( 14.2) | 77 ( 13.1) | 0.422 |
| Yes | 489 ( 85.5) | 480 ( 88.6) | 556 ( 85.8) | 511 ( 86.9) | ||
| Octreotide Acetate (%) | No | 557 ( 97.4) | 522 ( 96.3) | 619 ( 95.5) | 559 ( 95.1) | 0.194 |
| Yes | 15 (2.6) | 20 (3.7) | 29 (4.5) | 29 (4.9) | ||
| Vasopressin (%) | No | 375 ( 65.6) | 332 ( 61.3) | 341 ( 52.6) | 263 ( 44.7) | < 0.001 |
| Yes | 197 ( 34.4) | 210 ( 38.7) | 307 ( 47.4) | 325 ( 55.3) | ||
| Extracorporeal membrane oxygenation [ECMO] (%) | No | 571 ( 99.8) | 542 (100.0) | 648 (100.0) | 588 (100.0) | 0.375 |
| Yes | 1 (0.2) | 0 (0.0) | 0 (0.0) | 0 (0.0) | ||
| Non-invasive mechanical ventilation (%) | No | 570 ( 99.7) | 540 ( 99.6) | 648 (100.0) | 586 ( 99.7) | 0.513 |
| Yes | 2 (0.3) | 2 (0.4) | 0 (0.0) | 2 (0.3) | ||
| Heart failure (%) | No | 564 ( 98.6) | 537 ( 99.1) | 639 ( 98.6) | 579 ( 98.5) | 0.823 |
| Yes | 8 (1.4) | 5 (0.9) | 9 (1.4) | 9 (1.5) | ||
| Respiratory failure (%) | No | 452 ( 79.0) | 421 ( 77.7) | 492 ( 75.9) | 448 ( 76.2) | 0.556 |
| Yes | 120 ( 21.0) | 121 ( 22.3) | 156 ( 24.1) | 140 ( 23.8) | ||
| hepatopathy (%) | No | 532 ( 93.0) | 480 ( 88.6) | 556 ( 85.8) | 496 ( 84.4) | < 0.001 |
| Yes | 40 (7.0) | 62 ( 11.4) | 92 ( 14.2) | 92 ( 15.6) | ||
| nephropathy (%) | No | 529 ( 92.5) | 425 ( 78.4) | 409 ( 63.1) | 215 ( 36.6) | < 0.001 |
| Yes | 43 (7.5) | 117 ( 21.6) | 239 ( 36.9) | 373 ( 63.4) | ||
| dialysis_present (%) | No | 564 ( 98.6) | 526 ( 97.0) | 593 ( 91.5) | 402 ( 68.4) | < 0.001 |
| Yes | 8 (1.4) | 16 (3.0) | 55 (8.5) | 186 ( 31.6) | ||
The 2,350 patients were divided into two groups: survivors (1,530, 65.1%) and non-survivors (820, 34.9%) (Table S1). Compared to non-survivors, survivors had higher platelet count, hemoglobin, albumin, MAP, SBP, DBP, BT, and were more likely to use aspirin and heparin. Non-survivors demonstrated significantly higher levels of creatinine, BUN, TB, K, anion gap, CAR, HR, RR, SAPS II, SOFA, the percentage of patients used octreotide acetate, vasopressin, dialysis, suffering from RF, and hepatopathy was higher. Additionally, survivors exhibited longer durations of hospitalization stays (p < 0.05).
K-M survival curves
The K-M survival curves stratified by serum CAR quartiles, shown in Fig. 2, revealed significant differences in 28-day all-cause mortality among older adults patients with sepsis (p < 0.0001). Patients in the Q4 group had a significantly higher mortality rate compared to the other quartiles, indicating that serum CAR was strongly associated with 28-day mortality in the ICU.
Fig. 2.
K-M survival analysis curves for ICU 28-day all-cause mortality in older adults patients with sepsis
Associations of the CAR with 28-day all-cause mortality
The multivariable logistic regression analysis of the association between serum CAR and ICU 28-day all-cause mortality in elder patients with sepsis was shown in Table 2. After adjusting for covariates, in Model 1 (OR = 3.142, 95% CI: 2.439–4.064, p < 0.001) and Model 2 (OR = 1.640, 95% CI: 1.096–2.455, p < 0.05), the quartiles of CAR, especially the Q4 quartile, significantly associated with the 28-day all-cause mortality in the ICU of older adults patients with sepsis. The detailed results are presented in Table S2.
Table 2.
Associations of CAR with 28-day all-cause ICU mortality in elder patients with sepsis
| Model 1 | Model 2 | |||||||
|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | p | OR | 95% CI | p | |||
| groupQ1 | ref | ref | ref | ref | ref | ref | ref | ref |
| groupQ2 | 1.547 | 1.183 | 2.026 | 0.001 | 1.332 | 0.977 | 1.819 | 0.070 |
| groupQ3 | 2.187 | 1.701 | 2.821 | < 0.001 | 1.389 | 1.010 | 1.913 | 0.044 |
| groupQ4 | 3.142 | 2.439 | 4.064 | < 0.001 | 1.640 | 1.096 | 2.455 | 0.016 |
To assess the linear association between serum CAR and 28-day all-cause mortality in elder patients with sepsis, RCS analysis was employed to further investigate this relationship. The results indicated a significant association between serum CAR and 28-day all-cause mortality in elder patients with sepsis (p-overall < 0.001), and there was a nonlinear relationship (p-non-linear < 0.001) (Fig. 3).
Fig. 3.
Restricted cubic spline regression analysis for ICU 28-day all-cause mortality
Model 1: unadjusted. Model 2: adjusted for age, BMI, WBC, platelet count, hemoglobin, BUN, TB, Na, Ca, K, anion gap, MAP, HR, RR, SBP, DBP, BT, SAPS II, SOFA, aspirin, atorvastatin, heparin, octreotide acetate, vasopressin, non-invasive mechanical ventilation, HF, RF, hepatopathy, nephropathy, and dialysis presence.
Subgroup analysis
The forest plot of the subgroup analysis of older adults patients with sepsis was shown in Fig. 4, with detailed data presented in Table S3. In the subgroup analyses based on CAR quartiles for the categorical variable, the p-values for the interactions between each factor and serum CAR were all above 0.05, indicating no significant interactions. In addition, in subgroup analyses of the factors, the OR for race (black, other, white), gender (female and male), age (65–80 years and > 80 years), and BMI (Q2-Q4) were significantly higher than the reference group (Q1) (p < 0.05), which suggested a significant relationship between serum CAR and the risk of death in older adults patients with sepsis in these subgroups. Similarly, the use of medications such as aspirin, atorvastatin, octreotide acetate, vasopressin, and heparin, as well as non-dialysis status, were also significantly associated with the risk of death in older adults patients with sepsis.
Fig. 4.
Forest plot of subgroup analysis. P for interaction indicates the P-value of the interaction. When the P-value > 0.05, it indicates that there is no significant interaction between the factor and CAR, while when the P-value < 0.05, it indicates that there is a significant interaction
Discussion
Our research analyzed 2,350 older adults patients with sepsis and revealed that the CAR, as a biomarker comprehensively reflecting renal function and nutritional status, was significantly positively associated with 28-day all-cause mortality in the ICU. This association remained consistent across multivariate logistic regression analysis, restricted cubic spline analysis, and subgroup analyses. The findings suggested that CAR might integrate key pathophysiological mechanisms in the sepsis process, specifically the interaction between renal dysfunction and nutritional status. Compared with other prognostic scoring systems (e.g., SOFA, SAPS II), CAR offers advantages in simplicity and immediacy, allowing for rapid bedside acquisition without complex calculations, which is particularly applicable to resource-limited medical environments.
Baseline characteristics showed that as CAR quartiles increased, patients exhibited more severe organ dysfunction (e.g., higher SOFA and SAPS II scores), greater inflammatory burden (elevated WBC), and more pronounced metabolic derangements (increased serum potassium and anion gap, decreased albumin and blood pressure) [22]. Additionally, higher proportions of patients in higher CAR quartiles received vasopressor therapy and dialysis, suggesting that increased CAR may represent a cascade of worsening sepsis, where renal dysfunction and hypoproteinemia jointly exacerbated the risk of microcirculatory dysfunction and organ failure [23]. Notably, the non-survivor group had higher proportions of respiratory failure and hepatopathy, as well as greater use of octreotide acetate, vasopressin, and dialysis, findings likely related to more severe illness, higher complication burden, poor drug tolerance, and clinicians’ tendencies to employ multiple therapies in critical cases [24]. Additionally, patients in the highest CAR quartile and survivors had significantly longer hospital stays, whereas non-survivors exhibited relatively shorter stays due to rapid clinical deterioration and death, suggesting that CAR may reflect not only mortality risk but also disease severity and recovery trajectory [11]. The highest 28-day all-cause mortality in ICU observed in the survival curve for the highest CAR quartile further corroborated this association.
The results of multivariate logistic regression analysis aligned with previous studies on the impact of AKI severity and hypoproteinemia on sepsis prognosis [25]. For example, a study involving 1,123 ICU sepsis patients, the ICU mortality rate was 31.9%, and the 30-day mortality rate was 28.1%. Low serum albumin-to-creatinine ratio (ACR) levels were identified as independent risk factors associated with increased in-hospital, ICU, and 30-day mortality in sepsis patients [26]. This study further revealed through nonlinear analysis that the association between CAR and ICU 28-day all-cause mortality exhibited a steep upward trend in the high-value range, suggesting a critical threshold effect. However, these findings differed from a study in community-acquired pneumonia patients, which showed that albumin levels alone had a stronger association with mortality than creatinine, likely related to differences in disease type (pneumonia vs. sepsis) and the higher prevalence of AKI in ICU patients, indicating that the prognostic value of CAR may be disease-specific and warrants further validation across different clinical scenarios [27].
Additionally, the nonlinear relationship between CAR and ICU 28-day all-cause mortality, distinct from the linear associations of individual creatinine or albumin indices, may be attributed to the ratio’s ability to more precisely reflect the dynamic imbalance between catabolism and anabolism in sepsis. For example, in the early stage of sepsis, excessive inflammatory responses lead to albumin extravasation into the extravascular space, causing CAR to rise even before creatinine significantly increases, which signals potential risk of poor prognosis [28]. In the later stage, the rapid increase in creatinine and sustained depletion of albumin together exacerbate internal environmental disorders, forming a vicious cycle. Subgroup analyses showed that the prognostic value of CAR remained consistent across stratifications of race, gender, age, and BMI, and was not altered by the protective effects of aspirin or heparin, nor by treatments such as vasopressor use or dialysis. These findings further support CAR as a universal predictor of prognosis in older adults sepsis populations, with a mechanism potentially independent of traditional demographic factors and directly reflective of the core pathophysiological processes of sepsis [29].
The findings of this study hold significant clinical implications, providing a simple and effective early risk stratification tool for older adults patients with sepsis [30]. Elevated CAR not only indicates renal dysfunction but also reflects the body’s status under sepsis stress, enabling clinicians to identify high-risk patients early, promptly formulate and adjust clinical decisions, such as enhancing organ function support (e.g., early goal-directed therapy, renal replacement therapy) and nutritional interventions (e.g., high-protein enteral nutrition), thereby optimizing resource allocation and implementing targeted interventions.
Nonetheless, this study has several limitations. First, the database source had regional and cultural biases, as MIMIC database data primarily originated from Europe and North America with a relatively concentrated patient population. Such circumstances may lead to differences in medical practices (e.g., treatment protocols, testing frequency) and disease profiles (e.g., race-specific disease pathophysiology), potentially affecting the generalizability of findings to other global regions. Second, the retrospective design posed data quality risks, with challenges in ensuring complete data integrity and accuracy, such as missing key variables in electronic medical records, inconsistencies in measurement devices and methods, and biases in diagnostic coding. Although data were processed accordingly, some residual selection bias may have remained, impacting result generalizability [31]. Many patients had multiple underlying comorbidities, while multivariate adjustments were performed, it was difficult to completely exclude the influence of potential confounding factors. However, although the Logistic regression results showed a significant association between high CAR and increased 28-day ICU all-cause mortality, there might be other unmeasured or undetected confounding factors. Therefore, there are certain limitations in determining the causal relationship, which restricts the extrapolation of the results to actual clinical practice and other populations [32].
Future research can focus on exploring the mechanism of action of CAR in different types of sepsis patients, further clarifying its predictive value and biological significance across populations stratified by infection source, age, and other characteristics. Meanwhile, multicenter database studies should be conducted to validate the applicability of CAR in different geographical regions and healthcare settings, thereby expanding its clinical utility. Additionally, integrating CAR with other biomarkers for comprehensive assessment to construct more holistic sepsis prognosis prediction models may enhance prediction accuracy and reliability. Based on CAR’s predictive ability, future research may also explore the development of early interventions for high-risk patients, such as albumin supplementation or modulation of creatinine metabolism, to improve outcomes. Additionally, prospective studies or randomized controlled trials need to be conducted to further validate the causal relationship and offer more instructive evidence-based guidance for clinical practice.
Conclusion
Our findings demonstrate that CAR can be readily incorporated into the routine assessment of older adults patients with sepsis admitted to the ICU, serving as a simple, cost-effective, and easily accessible biomarker to quickly identify patients at high risk of mortality within 28 days. This enables timely initiation of intensive monitoring, individualized treatment strategies, and resource allocation, potentially improving patient outcomes. Future research could focus on validating these findings in independent cohorts and exploring the underlying mechanisms by which CAR influences the prognosis of older adults patients with sepsis, aiming to develop targeted interventions to reduce mortality. This study not only deepens our understanding of the association between serum-based biomarkers and the prognosis of older adults patients with sepsis but also provides practical guidance for clinical decision-making and future research directions.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Acknowledgements
The present study was conducted using data from the MIMIC-IV database, and we extend our sincere thanks to all staff and patients who participated in the development of these resources.
Abbreviations
- Cr
Creatinine
- Alb
Albumin
- CAR
Creatinine to Albumin Ratio
- MIMIC-IV
Medical Information Mart for Intensive Care IV
- K-M
Kaplan-Meier
- RCS
Restricted Cubic Spline
- ICU
Intensive Care Unit
- BIDMC
Beth Israel Deaconess Medical Center
- IRB
Institutional Review Board
- BMI
Body Mass Index
- HR
Heart Rate
- RR
Respiratory Rate
- MAP
Mean Arterial Pressure
- SBP
Systolic Blood Pressure
- DBP
Diastolic Blood Pressure
- BT
Body Temperature
- WBC
White Blood Cell Count
- BUN
Blood Urea Nitrogen
- TB
Total Bilirubin
- K
Potassium
- Na
Sodium
- Ca
Calcium
- SAPS II
Simplified Acute Physiology Score II
- SOFA
Sequential Organ Failure Assessment
- AKI
Acute Kidney Injury
- RF
Respiratory Failure
- HF
Heart Failure
- AF
Atrial Fibrillation
- ECMO
Extracorporeal Membrane Oxygenation
- LOS
Length of Stay
- SD
Standard Deviation
- CI
Confidence Interval
- OR
Odds Ratios
Author contributions
L.Z.*¹,² conducted data extraction from MIMIC-IV databases, performed statistical analysis, and drafted the original manuscript. H.C.*¹ contributed to systematic literature review, validated data interpretation, and co-drafted the manuscript. S.Y.*³ developed data processing pipelines for standardized data cleaning, and contributed to statistical methodology design. F.C.¹ analyzed clinical variables, ensured data quality control through institutional review boards, and validated findings across datasets. C.Z.¹ organized multicenter datasets from three participating hospitals, implemented validation protocols, and ensured compliance with data privacy regulations. Y.L.¹# designed the study framework, supervised methodology execution across institutions, and revised the manuscript critically for scientific rigor. Y.F.C.¹# conceptualized the research hypothesis, coordinated collaboration between emergency medicine and critical care departments, and verified analytical outputs through blinded re-evaluation. *These authors contributed equally to this work. #Corresponding authors: Y.L. (yancunliu@tmu.edu.cn); Y.F.C. (chaiyanfen2012@126.com)
Funding
No external funding was received for this study. All costs related to data acquisition, software development, and manuscript preparation were covered through institutional resources.
Data availability
The data utilized in this study were obtained from the MIMIC-IV database, accessible at https://mimic.physionet.org/. The data in this research can be reasonably applied to the corresponding author.
Declarations
Ethics approval and consent to participate
This study utilized de-identified data from the MIMIC-IV database, which was approved by the Institutional Review Board (IRB) of the Massachusetts Institute of Technology (MIT). All original data collection procedures complied with ethical guidelines and regulations, and informed consent was obtained for the primary data collection. Due to the retrospective and de-identified nature of the data, the requirement for additional ethical approval or patient consent for this secondary analysis was waived by the MIT IRB.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Lixin Zhao, Haixia Chai and Sifang Yu contributed equally to this work and co-first authors.
Contributor Information
Yancun Liu, Email: yancunliu@tmu.edu.cn.
Yanfen Chai, Email: chaiyanfen2012@126.com.
References
- 1.O’Brien JM Jr., Ali NA, Aberegg SK, Abraham E. Sepsis Am J Med. 2007;120(12):1012–22. [DOI] [PubMed] [Google Scholar]
- 2.Ramoni D, Tirandi A, Montecucco F, Liberale L. Sepsis in elderly patients: the role of neutrophils in pathophysiology and therapy. Intern Emerg Med. 2024;19(4):901–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Srdić T, Đurašević S, Lakić I, Ružičić A, Vujović P, Jevđović T et al. From molecular mechanisms to clinical therapy: Understanding Sepsis-Induced multiple organ dysfunction. Int J Mol Sci. 2024;25(14). [DOI] [PMC free article] [PubMed]
- 4.Pregernig A, Müller M, Held U, Beck-Schimmer B. Prediction of mortality in adult patients with sepsis using six biomarkers: a systematic review and meta-analysis. Ann Intensive Care. 2019;9(1):125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ronco C, Bellomo R, Kellum JA. Acute kidney injury. Lancet. 2019;394(10212):1949–64. [DOI] [PubMed] [Google Scholar]
- 6.White KC, Serpa-Neto A, Hurford R, Clement P, Laupland KB, See E, et al. Sepsis-associated acute kidney injury in the intensive care unit: incidence, patient characteristics, timing, trajectory, treatment, and associated outcomes. A multicenter, observational study. Intensive Care Med. 2023;49(9):1079–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gustot T, Durand F, Lebrec D, Vincent JL, Moreau R. Severe sepsis in cirrhosis. Hepatology. 2009;50(6):2022–33. [DOI] [PubMed] [Google Scholar]
- 8.Hu W, Chen H, Ma C, Sun Q, Yang M, Wang H, et al. Identification of indications for albumin administration in septic patients with liver cirrhosis. Crit Care. 2023;27(1):300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Grams ME, Coresh J, Matsushita K, Ballew SH, Sang Y, Surapaneni A, et al. Estimated glomerular filtration rate, albuminuria, and adverse outcomes: an Individual-Participant data Meta-Analysis. JAMA. 2023;330(13):1266–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Tao J, Sang D, Zhen L, Zhang X, Li Y, Wang G, et al. Elevated urine albumin-to-creatinine ratio increases the risk of new-onset heart failure in patients with type 2 diabetes. Cardiovasc Diabetol. 2023;22(1):70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Wang J, Li H, Luo H, Shi R, Chen S, Hu J, et al. Association between serum creatinine to albumin ratio and short- and long-term all-cause mortality in patients with acute pancreatitis admitted to the intensive care unit: a retrospective analysis based on the MIMIC-IV database. Front Immunol. 2024;15:1373371. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Schultes B, Emmerich S, Kistler AD, Mecheri B, Schnell O, Rudofsky G. Impact of Albumin-to-Creatinine ratio Point-of-Care testing on the diagnosis and management of diabetic kidney disease. J Diabetes Sci Technol. 2023;17(2):428–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Li S, Xie X, Zeng X, Wang S, Lan J. Association between serum albumin to serum creatinine ratio and mortality risk in patients with heart failure. Clin Transl Sci. 2023;16(11):2345–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Johnson AEW, Bulgarelli L, Shen L, Gayles A, Shammout A, Horng S, et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci Data. 2023;10(1):1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Zheng R, Qian S, Shi Y, Lou C, Xu H, Pan J. Association between triglyceride-glucose index and in-hospital mortality in critically ill patients with sepsis: analysis of the MIMIC-IV database. Cardiovasc Diabetol. 2023;22(1):307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Larson MG. Analysis of variance. Circulation. 2008;117(1):115–21. [DOI] [PubMed] [Google Scholar]
- 17.Ranstam J, Cook JA. Kaplan-Meier curve. Br J Surg. 2017;104(4):442. [DOI] [PubMed] [Google Scholar]
- 18.Ye Z, An S, Gao Y, Xie E, Zhao X, Guo Z, et al. Association between the triglyceride glucose index and in-hospital and 1-year mortality in patients with chronic kidney disease and coronary artery disease in the intensive care unit. Cardiovasc Diabetol. 2023;22(1):110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tripepi G, Jager KJ, Dekker FW, Zoccali C. Linear and logistic regression analysis. Kidney Int. 2008;73(7):806–10. [DOI] [PubMed] [Google Scholar]
- 20.Wei X, Chen X, Zhang Z, Wei J, Hu B, Long N, et al. Risk analysis of the association between different hemoglobin glycation index and poor prognosis in critical patients with coronary heart disease-A study based on the MIMIC-IV database. Cardiovasc Diabetol. 2024;23(1):113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Williamson SF, Grayling MJ, Mander AP, Noor NM, Savage JS, Yap C, et al. Subgroup analyses in randomized controlled trials frequently categorized continuous subgroup information. J Clin Epidemiol. 2022;150:72–9. [DOI] [PubMed] [Google Scholar]
- 22.Faix JD. Biomarkers of sepsis. Crit Rev Clin Lab Sci. 2013;50(1):23–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Cohen J. The Immunopathogenesis of sepsis. Nature. 2002;420(6917):885–91. [DOI] [PubMed] [Google Scholar]
- 24.Sun X, Zhang X, Tang R, Tian J, Li Y, Hu X, et al. Advanced lung cancer inflammation index is associated with mortality in critically ill patients with heart failure. ESC Heart Fail. 2025;12(1):508–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Turkyilmaz E, Ozkalayci F, Birdal O, Karagoz A, Tanboga IH, Tanalp AC, et al. Serum albumin to creatinine ratio and Short-Term clinical outcomes in patients with ST-Elevation myocardial infarction. Angiology. 2022;73(9):809–17. [DOI] [PubMed] [Google Scholar]
- 26.Hu Z, Song C, Zhang J. Elevated serum albumin-to-creatinine ratio as a protective factor on clinical outcomes among critically ill patients with sepsis: a retrospective study. Front Med (Lausanne). 2024;11:1436533. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Poston JT, Koyner JL. Sepsis associated acute kidney injury. BMJ. 2019;364:k4891. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Dull RO, Hahn RG. Hypovolemia with peripheral edema: what is wrong? Crit Care. 2023;27(1):206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lin W, Fu C, Miao J, Hong W, Chen X, Yan S, et al. Association between the serum albumin-creatinine ratio and 28-day intensive care unit mortality among patients with sepsis: a multicenter retrospective cohort study. Front Med (Lausanne). 2024;11:1484370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Kong S, Yu S, He W, He Y, Chen W, Zhang Y, et al. Serum Albumin-to-Creatinine ratio: A novel predictor of pulmonary infection in patients with ST-Segment elevation myocardial infarction undergoing percutaneous coronary intervention. J Atheroscler Thromb. 2024;31(12):1680–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Guan C, Gong A, Zhao Y, Yin C, Geng L, Liu L, et al. Interpretable machine learning model for new-onset atrial fibrillation prediction in critically ill patients: a multi-center study. Crit Care. 2024;28(1):349. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Cai W, Xu J, Wu X, Chen Z, Zeng L, Song X, et al. Association between triglyceride-glucose index and all-cause mortality in critically ill patients with ischemic stroke: analysis of the MIMIC-IV database. Cardiovasc Diabetol. 2023;22(1):138. [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.
Supplementary Materials
Data Availability Statement
The data utilized in this study were obtained from the MIMIC-IV database, accessible at https://mimic.physionet.org/. The data in this research can be reasonably applied to the corresponding author.