Creatinine-to-albumin ratio and all-cause mortality in patients with upper gastrointestinal bleeding: a retrospective study Authors: Shuohui Li, Xiaoli Xie, Weiwei Niu, Jiajing Xing, Kun Lei, Feng Gao, Fanhao Lu, Hongjing Zhao, Shuxian Cheng, Na Wang Published: 2025 ABSTRACT BackgroundThis research aimed to assess the relationship between creatinine-to-albumin ratio (CAR) and clinical outcomes among individuals with upper gastrointestinal bleeding (UGIB), with a particular focus on the association between CAR and 30-day and 90-day all-cause mortality (ACM).MethodsPatient data were obtained from the Medical Information Mart for Intensive Care IV version 3.1 (MIMIC-IV 3.1) and the intensive care unit of the Second Hospital of Hebei Medical University (HMUSH). The included individuals were stratified into four groups according to CAR quartiles. The 30-day ACM served as the primary outcome, and the 90-day ACM as the secondary outcome. Clinical outcomes across groups were compared by Kaplan–Meier curves. Cox proportional hazards (PH) regression and restricted cubic spline (RCS) models were applied to examine the association between CAR and mortality outcomes. Moreover, subgroup analyses were carried out.ResultsIn total, 1,685 individuals from the MIMIC-IV cohort and 408 individuals from the HMUSH cohort were enrolled. Kaplan–Meier curves demonstrated that higher CAR values were related to increased risks of 30-day and 90-day ACM (log-rank < 0.05). Cox PH regression indicated that those in the highest CAR quartile had 211% and 205% elevated risks of 30-day and 90-day ACM, respectively, in the MIMIC-IV cohort, and 241% and 182% higher risks, respectively, in the HMUSH cohort. Subgroup analysis demonstrated a significant interaction with concurrent cirrhosis.ConclusionElevated CAR values were significantly related to increased risks of 30-day and 90-day ACM among individuals with UGIB. These findings provide further insight into the relationship between serum indicators and prognosis among UGIB patients.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12876-025-04564-w. Introduction Upper gastrointestinal bleeding (UGIB) refers to acute hemorrhagic diseases originating above the ligament of Treitz and represents one of the most common emergencies in clinical practice [1]. UGIB patients typically present with hematemesis and melena, often accompanied by severe hypovolemia, anemia, and azotemia [2]. The mortality rates of UGIB range from 2% to 10% [3, 4]. The main causes of UGIB encompass peptic ulcers, esophageal varices, gastric cancer, and Mallory–Weiss syndrome [5, 6]. Despite advances in medical techniques and the widespread application of endoscopic hemostasis that have substantially improved cure rates [7], the risk of rebleeding among UGIB patients remains high [8], and patient outcomes vary considerably. Nevertheless, studies investigating prognostic indicators of severe UGIB are scarce. Early detection of patients at high risk and the implementation of individualized therapeutic strategies are critical for enhancing patients’ prognosis. Several studies have demonstrated that the prognostic performance of the creatinine-to-albumin ratio (CAR) exceeds that of either parameter alone [9], and CAR is related to the prognosis in multiple illnesses, including acute pancreatitis, sepsis, and hemorrhagic stroke [10–12]. Serum creatinine (Cr), an extensively applied indicator of renal impairment, reflects the severity of acute kidney injury (AKI) [13, 14]. Among individuals with UGIB, the incidence of AKI is reported to be 30%–40%, and both elevated serum Cr and the presence of AKI are strongly correlated with adverse outcomes [15–18]. Moreover, albumin (ALB) is an indicator of inflammatory activity and nutritional status. During stress states, redistribution of ALB and accelerated protein catabolism are often observed [19, 20]. Thus, CAR, by integrating serum Cr and ALB, reflects the dual pathophysiological processes of inflammation–nutrition imbalance and renal injury induced by UGIB. However, the potential of CAR in the prognosis of UGIB remains underexplored. Given its potential relevance in reflecting the pathobiology of UGIB, the present research aimed to estimate the relationship between CAR and clinical outcomes among individuals with UGIB, with a particular focus on the link between CAR and 30-day and 90-day all-cause mortality (ACM). Materials and methods Data sources One source of data for this research was the Medical Information Mart for Intensive Care IV version 3.1 (MIMIC-IV 3.1) [21, 22]. This database recorded clinical data on more than 190,000 patients and 450,000 inpatients between 2008 and 2022, including demographics, medications, laboratory examinations, medication administration, surgical procedures, vital signs, diagnoses, as well as follow-up survival outcomes. The training on the protection of human research participants designed by the National Institutes of Health was completed, and the Collaborative Institutional Training Initiative exam (Record ID 68429492) was passed to gain access to the data. As all included individuals in this database were anonymized, informed consent was waived. Data from individuals admitted to the intensive care unit (ICU) of the Second Hospital of Hebei Medical University (HMUSH) for UGIB between May 2023 and April 2025 were also collected. Demographics, laboratory and imaging results, treatment, and diagnoses during hospitalization were stored in a nonpublic electronic medical record system. The Ethics Committee of the HMUSH approved this study (Ethics ID: 2025-R653). Since all protected health information was de-identified, and clinical management was not affected, informed consent was waived. Study design and population Inclusion criteria encompassed: (i) individuals who were admitted to the ICU with UGIB and were diagnosed according to the International Classification of Diseases codes (ICD-9/10); (ii) older than 18 years; (iii) first ICU admission; and (iv) ICU stay of more than 24 h. Exclusion criteria were missing or abnormal serum ALB or Cr values within 24 h of admission (Fig. 1). Fig. 1Flowchart of the patient selection Flowchart of the patient selection Data extraction Data of the enrolled patients from the MIMIC-IV database were obtained via PostgreSQL 17.2 with SQL code. Patient data from the HMUSH were derived from the hospital electronic medical record system. Collected information encompassed: (i) demographics: sex and age; (ii) initial vital signs at ICU admission: diastolic blood pressure (DBP), heart rate (HR), systolic blood pressure (SBP), and respiratory rate (RR); (iii) comorbidities: diabetes mellitus (DM), hypertension (HTN), pulmonary infection, cirrhosis, and ascites; (iv) long-term medications: antiplatelet drugs, diuretics, calcium channel blockers (CCBs), beta-blockers; (v) treatments: ALB infusion, vasoactive drugs, antibacterial agents, and analgesics/sedatives; (vi) first laboratory test findings within 24 h of ICU admission: ALB, Cr, aspartate aminotransferase (AST), platelet count (PLT), total bilirubin (TBIL), hemoglobin (HB), alanine aminotransferase (ALT), white blood cell count (WBC), international normalized ratio (INR), alkaline phosphatase (ALP), potassium, sodium, chloride, magnesium, prothrombin time (PT), and blood urea nitrogen (BUN); (vii) length of hospital stay and survival outcomes: ICU stay, 30-day ACM, and 90-day ACM. Variables with missing data exceeding 20% were removed. For variables with less than 20% missing data, multiple imputation was performed using the MICE package in R, with 5 imputation iterations and a random seed set to 2025 to ensure reproducibility. Convergence diagnostics indicated that the imputation process had reached adequate convergence, and the imputed values were stable across iterations. (Table S1) Outcomes The 30-day ACM served as the primary endpoint, and the 90-day ACM as the secondary endpoint. Calculation of CAR CAR was the main exposure variable, calculated as Cr (µmol/L)/ALB (g/L) [10]. Statistical analysis Continuous variables were tested for normality. Median (interquartile range, IQR) was used to represent non-normally distributed continuous variables, and between-group comparisons were made via the Mann–Whitney U test. Frequencies and percentages were used to represent categorical variables, and between-group comparisons were made by Fisher’s exact test or Pearson’s Chi-square test. The enrolled individuals were grouped according to the quartiles of CAR. The relationship between CAR and outcomes was examined by Kaplan–Meier curves with the log-rank test. Multicollinearity among all variables included in the models was examined, and variables with a variance inflation factor ≥ 5 were removed to reduce collinearity (Tables S2-S3). Cox proportional hazards models were utilized to examine the association between CAR and clinical outcomes. Model 1 was unadjusted. Model 2 was adjusted for sex and age. Model 3 was adjusted for sex, age, HR, SBP, DBP, RR, DM, HTN, cirrhosis, pulmonary infection, ascites, use of antiplatelet drugs, ALB infusion, antibiotic use, HB, PLT, AST, and INR. Potential nonlinear associations between CAR and ACM were examined through restricted cubic spline (RCS) analyses. The knot locations for the RCS were set at 0.05, 0.35, 0.65, and 0.95. The reference value for the spline was set at the median value of the variable. Additionally, we included all relevant covariates in the model to account for potential confounders and ensure an accurate estimation of the nonlinear association. Subgroup analyses stratified by sex (female or male), age (≥ 65 or < 65 years), DM, HTN, cirrhosis, pulmonary infection, ascites, use of antiplatelet drugs, ALB infusion, and antibiotic use were carried out. R software (version 4.3.2) was employed to carry out statistical analyses. A P value < 0.05 was regarded as statistically significant. Results Baseline characteristics In total, 1,685 individuals from the MIMIC-IV cohort and 408 individuals from the HMUSH cohort were enrolled. Baseline characteristics were stratified by quartiles of CAR (Table 1). In the MIMIC-IV cohort, Q1 corresponded to CAR < 2.40, Q2 to 2.40 ≤ CAR < 3.64, Q3 to 3.64 ≤ CAR < 6.35, and Q4 to CAR ≥ 6.35. There were 644 female and 1,041 male patients. A total of 340 individuals (20.18%) died within 30 days, and 459 (27.24%) died within 90 days. In comparison to Q1, Q4 patients were older, had lower SBP, DBP, prevalence of HTN, ALB, HB, PLT, sodium, and chloride, but had higher RR, greater prevalence of DM and pulmonary infection, greater long-term administration of antiplatelet drugs, diuretics, CCBs, and β-blockers, more frequent ALB infusion, higher use of vasoactive drugs, antibacterial agents, and analgesic/sedatives, along with elevated Cr, WBC, ALP, AST, TBIL, BUN, potassium, magnesium, INR, and PT. In the HMUSH cohort, Q1 corresponded to CAR < 1.950, Q2 to 1.950 ≤ CAR < 2.585, Q3 to 2.585 ≤ CAR < 3.930, and Q4 to CAR ≥ 3.930. There were 285 female and 123 male patients. A total of 58 individuals (14.22%) died within 30 days, and 83 (20.34%) died within 90 days. Compared to Q1, Q4 patients had lower ALB and chloride but higher HR, RR, greater prevalence of DM and pulmonary infection, greater long-term administration of CCBs, more frequent ALB infusion, higher use of vasoactive drugs, antibacterial agents, and analgesic/sedatives, along with elevated Cr, WBC, BUN, sodium, potassium, and magnesium. Both cohorts consistently demonstrated that those with elevated CAR had longer ICU stays and greater risks of 30-day and 90-day ACM. Table 1Baseline characteristics of patients grouped according to CAR index quartilesMIMIC CohortHMUSH CohortCharacteristicN = 1,685Q1(CAR < 2.40,n = 420)Q2(2.40 ≤ CAR < 3.64,n = 421)Q3(3.64 ≤ CAR < 6.35,n = 421)Q4(CAR ≥ 6.35,n = 423) p N = 408Q1(CAR < 1.950,n = 102)Q2(1.950 ≤ CAR < 2.585,n = 102)Q3(2.585 ≤ CAR < 3.930,n = 102)Q4(CAR ≥ 3.930,n = 102)pGender, n(%)< 0.001< 0.001 Female644(38.22%)197(46.90%)161(38.24%)141(33.49%)145(34.28%)285(69.85%)47(46.08%)73(71.57%)87(85.29%)78(76.47%) Male1,041(61.78%)223(53.10%)260(61.76%)280(66.51%)278(65.72%)123(30.15%)55(53.92%)29(28.43%)15(14.71%)24(23.53%)Age(years)63.62(53.03,75.23)60.25(48.22,71.35)63.60(53.60,76.03)65.92(55.99,77.72)64.17(53.56,74.62)< 0.00164.00(53.50,72.00)60.50(54.00,70.00)60.50(52.00,72.00)66.00(58.00,73.00)65.00(50.00,75.00)0.095HR(beats/min)91.00(78.00,106.00)92.00(79.00,108.00)92.00(77.00,108.00)92.00(80.00,107.00)90.00(77.00,103.00)0.24586.00(77.00,99.00)84.00(75.00,95.00)83.50(75.00,96.00)87.50(77.00,100.00)90.00(79.00,109.00)0.038SBP(mmHg)119.00(103.00,137.00)124.00(108.00,140.00)119.00(106.00,135.00)117.00(101.00,136.00)115.00(99.00,137.00)< 0.001120.00(105.00,137.00)123.50(111.00,140.00)119.00(105.00,136.00)116.50(103.00,133.00)120.00(103.00,137.00)0.130DBP(mmHg)67.00(57.00,79.00)72.00(60.00,83.00)67.00(59.00,78.00)64.00(54.00,75.00)64.00(53.00,78.00)< 0.00169.00(60.00,79.00)70.50(63.00,81.00)68.50(60.00,80.00)67.00(59.00,74.00)69.00(60.00,79.00)0.171RR(times/min)19.00(16.00,23.00)19.00(16.00,22.00)19.00(16.00,22.00)20.00(16.00,24.00)20.00(16.00,24.00)< 0.00118.00(16.00,21.00)17.00(15.00,20.00)17.50(16.00,20.00)19.00(16.00,22.00)19.00(16.00,22.00)0.015Diabetes Mellitus, n(%)< 0.0010.011 No1,050(62.31%)301(71.67%)257(61.05%)251(59.62%)241(56.97%)301(73.77%)76(74.51%)87(85.29%)70(68.63%)68(66.67%) Yes635(37.69%)119(28.33%)164(38.95%)170(40.38%)182(43.03%)107(26.23%)26(25.49%)15(14.71%)32(31.37%)34(33.33%)Hypertension, n(%)< 0.0010.097 No846(50.21%)201(47.86%)170(40.38%)209(49.64%)266(62.88%)194(47.55%)50(49.02%)58(56.86%)41(40.20%)45(44.12%) Yes839(49.79%)219(52.14%)251(59.62%)212(50.36%)157(37.12%)214(52.45%)52(50.98%)44(43.14%)61(59.80%)57(55.88%)Pneumonia, n(%)0.007< 0.001 No1,108(65.76%)303(72.14%)272(64.61%)275(65.32%)258(60.99%)225(55.15%)58(56.86%)67(65.69%)63(61.76%)37(36.27%) Yes577(34.24%)117(27.86%)149(35.39%)146(34.68%)165(39.01%)183(44.85%)44(43.14%)35(34.31%)39(38.24%)65(63.73%)Liver Cirrhosis, n(%)0.8050.123 No1,222(72.52%)309(73.57%)299(71.02%)303(71.97%)311(73.52%)287(70.34%)65(63.73%)69(67.65%)73(71.57%)80(78.43%) Yes463(27.48%)111(26.43%)122(28.98%)118(28.03%)112(26.48%)121(29.66%)37(36.27%)33(32.35%)29(28.43%)22(21.57%)Ascites, n(%)0.3080.641 No1,302(77.27%)334(79.52%)332(78.86%)318(75.53%)318(75.18%)332(81.37%)81(79.41%)80(78.43%)85(83.33%)86(84.31%) Yes383(22.73%)86(20.48%)89(21.14%)103(24.47%)105(24.82%)76(18.63%)21(20.59%)22(21.57%)17(16.67%)16(15.69%)Antiplatelet Drugs, n(%)0.0080.160 No903(53.59%)254(60.48%)216(51.31%)224(53.21%)209(49.41%)309(75.74%)82(80.39%)79(77.45%)79(77.45%)69(67.65%) Yes782(46.41%)166(39.52%)205(48.69%)197(46.79%)214(50.59%)99(24.26%)20(19.61%)23(22.55%)23(22.55%)33(32.35%)Diuretic, n(%)< 0.0010.287 No422(25.04%)151(35.95%)89(21.14%)76(18.05%)106(25.06%)368(90.20%)93(91.18%)94(92.16%)94(92.16%)87(85.29%) Yes1,263(74.96%)269(64.05%)332(78.86%)345(81.95%)317(74.94%)40(9.80%)9(8.82%)8(7.84%)8(7.84%)15(14.71%)Calcium Channel Blockers, n(%)< 0.0010.015 No1,452(86.17%)388(92.38%)364(86.46%)359(85.27%)341(80.61%)316(77.45%)83(81.37%)86(84.31%)79(77.45%)68(66.67%) Yes233(13.83%)32(7.62%)57(13.54%)62(14.73%)82(19.39%)92(22.55%)19(18.63%)16(15.69%)23(22.55%)34(33.33%)Beta-blockers, n(%)0.0070.154 No602(35.73%)171(40.71%)124(29.45%)149(35.39%)158(37.35%)374(91.67%)94(92.16%)98(96.08%)93(91.18%)89(87.25%) Yes1,083(64.27%)249(59.29%)297(70.55%)272(64.61%)265(62.65%)34(8.33%)8(7.84%)4(3.92%)9(8.82%)13(12.75%)Albumin Infusion, n(%)< 0.0010.005 No1,173(69.61%)344(81.90%)323(76.72%)257(61.05%)249(58.87%)112(27.45%)36(35.29%)33(32.35%)28(27.45%)15(14.71%) Yes512(30.39%)76(18.10%)98(23.28%)164(38.95%)174(41.13%)296(72.55%)66(64.71%)69(67.65%)74(72.55%)87(85.29%)Vasoactive, n(%)< 0.001< 0.001 No974(57.80%)318(75.71%)259(61.52%)219(52.02%)178(42.08%)292(71.57%)88(86.27%)79(77.45%)74(72.55%)51(50.00%) Yes711(42.20%)102(24.29%)162(38.48%)202(47.98%)245(57.92%)116(28.43%)14(13.73%)23(22.55%)28(27.45%)51(50.00%)Antimicrobial Drugs, n(%)< 0.0010.008 No626(37.15%)219(52.14%)181(42.99%)116(27.55%)110(26.00%)114(27.94%)28(27.45%)39(38.24%)30(29.41%)17(16.67%) Yes1,059(62.85%)201(47.86%)240(57.01%)305(72.45%)313(74.00%)294(72.06%)74(72.55%)63(61.76%)72(70.59%)85(83.33%)Sedative, n(%)0.214< 0.001 No537(31.87%)137(32.62%)148(35.15%)120(28.50%)132(31.21%)257(62.99%)72(70.59%)73(71.57%)70(68.63%)42(41.18%) Yes1,148(68.13%)283(67.38%)273(64.85%)301(71.50%)291(68.79%)151(37.01%)30(29.41%)29(28.43%)32(31.37%)60(58.82%)Albumin(g/L)30.0(25.0,34.0)31.5(27.5,36.0)30.0(26.0,34.0)29.0(25.0,33.0)28.0(23.0,33.0)< 0.00131.05(27.50,34.25)33.00(30.80,36.60)32.35(29.00,34.80)30.25(27.10,32.10)28.15(24.10,31.50)< 0.001Creatinine(umol/L)106.08(70.72,185.64)61.88(53.04,70.72)88.40(70.72,97.24)132.60(114.92,167.96)318.24(221.00,459.68)< 0.00179.50(63.00,116.50)53.50(44.00,60.00)72.00(64.60,80.00)91.00(80.00,105.00)190.00(138.00,253.00)< 0.001CAR3.66(2.41,6.36)1.88(1.63,2.15)2.87(2.63,3.21)4.71(4.14,5.42)10.48(8.07,16.50)< 0.0012.59(1.96,3.94)1.64(1.34,1.83)2.24(2.13,2.39)3.05(2.75,3.46)6.64(4.88,9.94)< 0.001WBC(k/uL)10.70(7.10,15.70)8.85(6.00,13.10)10.50(6.80,15.00)12.00(8.00,16.80)11.80(7.60,17.00)< 0.0019.22(6.32,13.35)7.15(5.20,10.80)8.97(6.10,12.80)8.90(6.40,13.03)11.54(8.70,15.69)< 0.001HB(g/L)94.0(80.0,111.0)97.0(84.0,113.0)96.0(83.0,115.0)93.0(80.0,109.0)90.0(76.0,103.0)< 0.00177.00(64.00,97.00)84.50(65.00,99.00)79.00(66.00,95.00)74.00(64.00,91.00)73.50(59.00,104.00)0.132PLT(k/uL)148.00(90.00,220.00)162.00(94.00,244.00)154.00(97.00,226.00)136.00(84.00,205.00)141.00(80.00,215.00)0.005147.50(87.50,219.00)139.00(71.00,221.00)151.50(101.00,216.00)150.50(88.00,208.00)138.00(88.00,222.00)0.89ALP(IU/L)85.00(61.00,129.00)76.50(55.00,119.00)80.00(58.00,125.00)84.00(61.00,123.00)99.00(71.00,151.00)< 0.00168.55(51.60,91.00)72.40(54.30,94.40)65.40(52.00,96.20)65.15(51.70,85.00)70.60(50.50,92.00)0.523ALT(IU/L)29.00(17.00,61.00)27.00(17.00,48.50)29.00(17.00,64.00)31.00(18.00,71.00)31.00(16.00,69.00)0.07921.15(12.30,42.30)21.85(12.40,34.60)21.30(11.90,34.30)20.55(14.30,47.30)21.05(11.10,68.30)0.582AST(IU/L)49.00(26.00,117.00)41.50(24.00,83.00)47.00(25.00,109.00)56.00(30.00,158.00)59.00(26.00,137.00)< 0.00127.45(17.90,51.80)26.40(17.70,39.30)24.40(17.00,38.30)27.70(18.30,48.40)30.20(17.90,113.70)0.063TBIL(mg/L)11.0(5.0,31.0)10.0(5.0,23.5)10.0(5.0,25.0)14.0(7.0,41.0)13.0(4.0,43.0)< 0.00113.80(9.30,23.50)15.89(9.60,25.20)15.05(10.30,21.40)13.40(9.10,23.60)12.15(7.70,21.30)0.221BUN(mmol/L)4.66(2.83,8.15)2.66(1.83,3.83)3.66(2.66,5.49)5.99(4.33,8.82)9.48(6.66,14.14)< 0.0019.35(5.95,14.45)6.00(3.92,9.50)7.65(5.40,11.10)9.45(6.90,13.50)17.10(12.50,25.80)< 0.001Sodion(mmol/L)138.00(134.00,141.00)139.00(136.00,141.00)138.00(135.00,141.00)138.00(134.00,142.00)136.00(131.00,140.00)< 0.001139.10(136.15,142.05)139.45(136.40,141.40)137.85(135.90,140.30)139.15(136.00,141.90)140.45(137.10,145.00)< 0.001Potassium(mmol/L)4.20(3.70,4.80)3.90(3.60,4.30)4.20(3.80,4.60)4.40(3.80,4.90)4.50(3.90,5.20)< 0.0014.21(3.86,4.65)3.94(3.66,4.32)4.11(3.78,4.50)4.30(3.98,4.78)4.60(4.10,5.15)< 0.001Chloridion(mmol/L)104.00(99.00,108.00)105.00(101.00,109.00)105.00(100.00,109.00)104.00(99.00,109.00)100.00(95.00,107.00)< 0.001106.80(103.15,110.30)107.30(102.80,110.00)105.50(103.00,108.40)107.80(104.70,111.00)107.25(103.60,111.70)0.013Magnesium(mmol/L)0.95(0.85,1.10)0.90(0.80,1.00)0.95(0.85,1.05)1.00(0.85,1.10)1.05(0.90,1.20)< 0.0010.82(0.74,0.91)0.80(0.73,0.87)0.80(0.75,0.87)0.83(0.74,0.89)0.88(0.74,1.02)< 0.001INR1.40(1.20,1.90)1.30(1.10,1.60)1.40(1.20,1.70)1.60(1.30,2.00)1.60(1.30,2.10)< 0.0011.19(1.08,1.40)1.16(1.07,1.31)1.20(1.08,1.37)1.18(1.08,1.35)1.24(1.08,1.54)0.147PT(s)15.80(13.30,20.20)14.50(12.60,17.70)15.00(13.10,18.40)17.60(14.50,21.90)16.90(13.80,23.00)< 0.00113.25(12.00,15.70)13.00(12.00,14.70)13.25(12.00,15.50)13.20(12.10,15.20)13.70(12.00,17.10)0.252LOS in ICU2.87(1.79,6.00)2.43(1.65,4.40)2.51(1.70,5.04)3.19(1.92,6.63)3.77(2.09,7.94)< 0.0015.00(3.00,10.00)6.00(3.00,11.00)5.00(3.00,8.00)5.00(3.00,7.00)6.00(4.00,13.00)0.01730-day mortality, n(%)340(20.18%)47(11.19%)60(14.25%)105(24.94%)128(30.26%)< 0.00158(14.22%)8(7.84%)12(11.76%)13(12.75%)25(24.51%)0.00590-day mortality, n(%)459(27.24%)65(15.48%)86(20.43%)141(33.49%)167(39.48%)< 0.00183(20.34%)14(13.73%)17(16.67%)17(16.67%)35(34.31%)< 0.001HR Heart Rate, SBP Systolic Blood Pressure, DBP Diastolic Blood Pressure, RR Respiratory Rate, CAR Albumin/Creatinine, WBC White Blood Cell, HB Hemoglobin, PLT Platelet, ALP Alkaline Phosphatase, ALT Alanine Aminotransferase, AST Aspartate Aminotransferase; TBIL Total Bilirubin, BUN Blood Urea Nitrogen, INR International Normalized Ratio, PT Prothrombin Time Table 1Baseline characteristics of patients grouped according to CAR index quartilesMIMIC CohortHMUSH CohortCharacteristicN = 1,685Q1(CAR < 2.40,n = 420)Q2(2.40 ≤ CAR < 3.64,n = 421)Q3(3.64 ≤ CAR < 6.35,n = 421)Q4(CAR ≥ 6.35,n = 423) p N = 408Q1(CAR < 1.950,n = 102)Q2(1.950 ≤ CAR < 2.585,n = 102)Q3(2.585 ≤ CAR < 3.930,n = 102)Q4(CAR ≥ 3.930,n = 102)pGender, n(%)< 0.001< 0.001 Female644(38.22%)197(46.90%)161(38.24%)141(33.49%)145(34.28%)285(69.85%)47(46.08%)73(71.57%)87(85.29%)78(76.47%) Male1,041(61.78%)223(53.10%)260(61.76%)280(66.51%)278(65.72%)123(30.15%)55(53.92%)29(28.43%)15(14.71%)24(23.53%)Age(years)63.62(53.03,75.23)60.25(48.22,71.35)63.60(53.60,76.03)65.92(55.99,77.72)64.17(53.56,74.62)< 0.00164.00(53.50,72.00)60.50(54.00,70.00)60.50(52.00,72.00)66.00(58.00,73.00)65.00(50.00,75.00)0.095HR(beats/min)91.00(78.00,106.00)92.00(79.00,108.00)92.00(77.00,108.00)92.00(80.00,107.00)90.00(77.00,103.00)0.24586.00(77.00,99.00)84.00(75.00,95.00)83.50(75.00,96.00)87.50(77.00,100.00)90.00(79.00,109.00)0.038SBP(mmHg)119.00(103.00,137.00)124.00(108.00,140.00)119.00(106.00,135.00)117.00(101.00,136.00)115.00(99.00,137.00)< 0.001120.00(105.00,137.00)123.50(111.00,140.00)119.00(105.00,136.00)116.50(103.00,133.00)120.00(103.00,137.00)0.130DBP(mmHg)67.00(57.00,79.00)72.00(60.00,83.00)67.00(59.00,78.00)64.00(54.00,75.00)64.00(53.00,78.00)< 0.00169.00(60.00,79.00)70.50(63.00,81.00)68.50(60.00,80.00)67.00(59.00,74.00)69.00(60.00,79.00)0.171RR(times/min)19.00(16.00,23.00)19.00(16.00,22.00)19.00(16.00,22.00)20.00(16.00,24.00)20.00(16.00,24.00)< 0.00118.00(16.00,21.00)17.00(15.00,20.00)17.50(16.00,20.00)19.00(16.00,22.00)19.00(16.00,22.00)0.015Diabetes Mellitus, n(%)< 0.0010.011 No1,050(62.31%)301(71.67%)257(61.05%)251(59.62%)241(56.97%)301(73.77%)76(74.51%)87(85.29%)70(68.63%)68(66.67%) Yes635(37.69%)119(28.33%)164(38.95%)170(40.38%)182(43.03%)107(26.23%)26(25.49%)15(14.71%)32(31.37%)34(33.33%)Hypertension, n(%)< 0.0010.097 No846(50.21%)201(47.86%)170(40.38%)209(49.64%)266(62.88%)194(47.55%)50(49.02%)58(56.86%)41(40.20%)45(44.12%) Yes839(49.79%)219(52.14%)251(59.62%)212(50.36%)157(37.12%)214(52.45%)52(50.98%)44(43.14%)61(59.80%)57(55.88%)Pneumonia, n(%)0.007< 0.001 No1,108(65.76%)303(72.14%)272(64.61%)275(65.32%)258(60.99%)225(55.15%)58(56.86%)67(65.69%)63(61.76%)37(36.27%) Yes577(34.24%)117(27.86%)149(35.39%)146(34.68%)165(39.01%)183(44.85%)44(43.14%)35(34.31%)39(38.24%)65(63.73%)Liver Cirrhosis, n(%)0.8050.123 No1,222(72.52%)309(73.57%)299(71.02%)303(71.97%)311(73.52%)287(70.34%)65(63.73%)69(67.65%)73(71.57%)80(78.43%) Yes463(27.48%)111(26.43%)122(28.98%)118(28.03%)112(26.48%)121(29.66%)37(36.27%)33(32.35%)29(28.43%)22(21.57%)Ascites, n(%)0.3080.641 No1,302(77.27%)334(79.52%)332(78.86%)318(75.53%)318(75.18%)332(81.37%)81(79.41%)80(78.43%)85(83.33%)86(84.31%) Yes383(22.73%)86(20.48%)89(21.14%)103(24.47%)105(24.82%)76(18.63%)21(20.59%)22(21.57%)17(16.67%)16(15.69%)Antiplatelet Drugs, n(%)0.0080.160 No903(53.59%)254(60.48%)216(51.31%)224(53.21%)209(49.41%)309(75.74%)82(80.39%)79(77.45%)79(77.45%)69(67.65%) Yes782(46.41%)166(39.52%)205(48.69%)197(46.79%)214(50.59%)99(24.26%)20(19.61%)23(22.55%)23(22.55%)33(32.35%)Diuretic, n(%)< 0.0010.287 No422(25.04%)151(35.95%)89(21.14%)76(18.05%)106(25.06%)368(90.20%)93(91.18%)94(92.16%)94(92.16%)87(85.29%) Yes1,263(74.96%)269(64.05%)332(78.86%)345(81.95%)317(74.94%)40(9.80%)9(8.82%)8(7.84%)8(7.84%)15(14.71%)Calcium Channel Blockers, n(%)< 0.0010.015 No1,452(86.17%)388(92.38%)364(86.46%)359(85.27%)341(80.61%)316(77.45%)83(81.37%)86(84.31%)79(77.45%)68(66.67%) Yes233(13.83%)32(7.62%)57(13.54%)62(14.73%)82(19.39%)92(22.55%)19(18.63%)16(15.69%)23(22.55%)34(33.33%)Beta-blockers, n(%)0.0070.154 No602(35.73%)171(40.71%)124(29.45%)149(35.39%)158(37.35%)374(91.67%)94(92.16%)98(96.08%)93(91.18%)89(87.25%) Yes1,083(64.27%)249(59.29%)297(70.55%)272(64.61%)265(62.65%)34(8.33%)8(7.84%)4(3.92%)9(8.82%)13(12.75%)Albumin Infusion, n(%)< 0.0010.005 No1,173(69.61%)344(81.90%)323(76.72%)257(61.05%)249(58.87%)112(27.45%)36(35.29%)33(32.35%)28(27.45%)15(14.71%) Yes512(30.39%)76(18.10%)98(23.28%)164(38.95%)174(41.13%)296(72.55%)66(64.71%)69(67.65%)74(72.55%)87(85.29%)Vasoactive, n(%)< 0.001< 0.001 No974(57.80%)318(75.71%)259(61.52%)219(52.02%)178(42.08%)292(71.57%)88(86.27%)79(77.45%)74(72.55%)51(50.00%) Yes711(42.20%)102(24.29%)162(38.48%)202(47.98%)245(57.92%)116(28.43%)14(13.73%)23(22.55%)28(27.45%)51(50.00%)Antimicrobial Drugs, n(%)< 0.0010.008 No626(37.15%)219(52.14%)181(42.99%)116(27.55%)110(26.00%)114(27.94%)28(27.45%)39(38.24%)30(29.41%)17(16.67%) Yes1,059(62.85%)201(47.86%)240(57.01%)305(72.45%)313(74.00%)294(72.06%)74(72.55%)63(61.76%)72(70.59%)85(83.33%)Sedative, n(%)0.214< 0.001 No537(31.87%)137(32.62%)148(35.15%)120(28.50%)132(31.21%)257(62.99%)72(70.59%)73(71.57%)70(68.63%)42(41.18%) Yes1,148(68.13%)283(67.38%)273(64.85%)301(71.50%)291(68.79%)151(37.01%)30(29.41%)29(28.43%)32(31.37%)60(58.82%)Albumin(g/L)30.0(25.0,34.0)31.5(27.5,36.0)30.0(26.0,34.0)29.0(25.0,33.0)28.0(23.0,33.0)< 0.00131.05(27.50,34.25)33.00(30.80,36.60)32.35(29.00,34.80)30.25(27.10,32.10)28.15(24.10,31.50)< 0.001Creatinine(umol/L)106.08(70.72,185.64)61.88(53.04,70.72)88.40(70.72,97.24)132.60(114.92,167.96)318.24(221.00,459.68)< 0.00179.50(63.00,116.50)53.50(44.00,60.00)72.00(64.60,80.00)91.00(80.00,105.00)190.00(138.00,253.00)< 0.001CAR3.66(2.41,6.36)1.88(1.63,2.15)2.87(2.63,3.21)4.71(4.14,5.42)10.48(8.07,16.50)< 0.0012.59(1.96,3.94)1.64(1.34,1.83)2.24(2.13,2.39)3.05(2.75,3.46)6.64(4.88,9.94)< 0.001WBC(k/uL)10.70(7.10,15.70)8.85(6.00,13.10)10.50(6.80,15.00)12.00(8.00,16.80)11.80(7.60,17.00)< 0.0019.22(6.32,13.35)7.15(5.20,10.80)8.97(6.10,12.80)8.90(6.40,13.03)11.54(8.70,15.69)< 0.001HB(g/L)94.0(80.0,111.0)97.0(84.0,113.0)96.0(83.0,115.0)93.0(80.0,109.0)90.0(76.0,103.0)< 0.00177.00(64.00,97.00)84.50(65.00,99.00)79.00(66.00,95.00)74.00(64.00,91.00)73.50(59.00,104.00)0.132PLT(k/uL)148.00(90.00,220.00)162.00(94.00,244.00)154.00(97.00,226.00)136.00(84.00,205.00)141.00(80.00,215.00)0.005147.50(87.50,219.00)139.00(71.00,221.00)151.50(101.00,216.00)150.50(88.00,208.00)138.00(88.00,222.00)0.89ALP(IU/L)85.00(61.00,129.00)76.50(55.00,119.00)80.00(58.00,125.00)84.00(61.00,123.00)99.00(71.00,151.00)< 0.00168.55(51.60,91.00)72.40(54.30,94.40)65.40(52.00,96.20)65.15(51.70,85.00)70.60(50.50,92.00)0.523ALT(IU/L)29.00(17.00,61.00)27.00(17.00,48.50)29.00(17.00,64.00)31.00(18.00,71.00)31.00(16.00,69.00)0.07921.15(12.30,42.30)21.85(12.40,34.60)21.30(11.90,34.30)20.55(14.30,47.30)21.05(11.10,68.30)0.582AST(IU/L)49.00(26.00,117.00)41.50(24.00,83.00)47.00(25.00,109.00)56.00(30.00,158.00)59.00(26.00,137.00)< 0.00127.45(17.90,51.80)26.40(17.70,39.30)24.40(17.00,38.30)27.70(18.30,48.40)30.20(17.90,113.70)0.063TBIL(mg/L)11.0(5.0,31.0)10.0(5.0,23.5)10.0(5.0,25.0)14.0(7.0,41.0)13.0(4.0,43.0)< 0.00113.80(9.30,23.50)15.89(9.60,25.20)15.05(10.30,21.40)13.40(9.10,23.60)12.15(7.70,21.30)0.221BUN(mmol/L)4.66(2.83,8.15)2.66(1.83,3.83)3.66(2.66,5.49)5.99(4.33,8.82)9.48(6.66,14.14)< 0.0019.35(5.95,14.45)6.00(3.92,9.50)7.65(5.40,11.10)9.45(6.90,13.50)17.10(12.50,25.80)< 0.001Sodion(mmol/L)138.00(134.00,141.00)139.00(136.00,141.00)138.00(135.00,141.00)138.00(134.00,142.00)136.00(131.00,140.00)< 0.001139.10(136.15,142.05)139.45(136.40,141.40)137.85(135.90,140.30)139.15(136.00,141.90)140.45(137.10,145.00)< 0.001Potassium(mmol/L)4.20(3.70,4.80)3.90(3.60,4.30)4.20(3.80,4.60)4.40(3.80,4.90)4.50(3.90,5.20)< 0.0014.21(3.86,4.65)3.94(3.66,4.32)4.11(3.78,4.50)4.30(3.98,4.78)4.60(4.10,5.15)< 0.001Chloridion(mmol/L)104.00(99.00,108.00)105.00(101.00,109.00)105.00(100.00,109.00)104.00(99.00,109.00)100.00(95.00,107.00)< 0.001106.80(103.15,110.30)107.30(102.80,110.00)105.50(103.00,108.40)107.80(104.70,111.00)107.25(103.60,111.70)0.013Magnesium(mmol/L)0.95(0.85,1.10)0.90(0.80,1.00)0.95(0.85,1.05)1.00(0.85,1.10)1.05(0.90,1.20)< 0.0010.82(0.74,0.91)0.80(0.73,0.87)0.80(0.75,0.87)0.83(0.74,0.89)0.88(0.74,1.02)< 0.001INR1.40(1.20,1.90)1.30(1.10,1.60)1.40(1.20,1.70)1.60(1.30,2.00)1.60(1.30,2.10)< 0.0011.19(1.08,1.40)1.16(1.07,1.31)1.20(1.08,1.37)1.18(1.08,1.35)1.24(1.08,1.54)0.147PT(s)15.80(13.30,20.20)14.50(12.60,17.70)15.00(13.10,18.40)17.60(14.50,21.90)16.90(13.80,23.00)< 0.00113.25(12.00,15.70)13.00(12.00,14.70)13.25(12.00,15.50)13.20(12.10,15.20)13.70(12.00,17.10)0.252LOS in ICU2.87(1.79,6.00)2.43(1.65,4.40)2.51(1.70,5.04)3.19(1.92,6.63)3.77(2.09,7.94)< 0.0015.00(3.00,10.00)6.00(3.00,11.00)5.00(3.00,8.00)5.00(3.00,7.00)6.00(4.00,13.00)0.01730-day mortality, n(%)340(20.18%)47(11.19%)60(14.25%)105(24.94%)128(30.26%)< 0.00158(14.22%)8(7.84%)12(11.76%)13(12.75%)25(24.51%)0.00590-day mortality, n(%)459(27.24%)65(15.48%)86(20.43%)141(33.49%)167(39.48%)< 0.00183(20.34%)14(13.73%)17(16.67%)17(16.67%)35(34.31%)< 0.001HR Heart Rate, SBP Systolic Blood Pressure, DBP Diastolic Blood Pressure, RR Respiratory Rate, CAR Albumin/Creatinine, WBC White Blood Cell, HB Hemoglobin, PLT Platelet, ALP Alkaline Phosphatase, ALT Alanine Aminotransferase, AST Aspartate Aminotransferase; TBIL Total Bilirubin, BUN Blood Urea Nitrogen, INR International Normalized Ratio, PT Prothrombin Time Baseline characteristics of patients grouped according to CAR index quartiles HR Heart Rate, SBP Systolic Blood Pressure, DBP Diastolic Blood Pressure, RR Respiratory Rate, CAR Albumin/Creatinine, WBC White Blood Cell, HB Hemoglobin, PLT Platelet, ALP Alkaline Phosphatase, ALT Alanine Aminotransferase, AST Aspartate Aminotransferase; TBIL Total Bilirubin, BUN Blood Urea Nitrogen, INR International Normalized Ratio, PT Prothrombin Time Kaplan-Meier survival curves Survival analyses for both cohorts indicated that in comparison to Q1, Q4 patients had significantly lower 30-day and 90-day survival (log-rank P < 0.05) (Fig. 2). Fig. 2Kaplan-Meier (K-M) survival analysis for ACM in individuals with UGIB. K–M curves for 30-day (A) and 90-day (B) ACM in the MIMIC-IV cohort; K–M curves for 30-day (C) and 90-day (D) ACM in the HBUSH cohort Kaplan-Meier (K-M) survival analysis for ACM in individuals with UGIB. K–M curves for 30-day (A) and 90-day (B) ACM in the MIMIC-IV cohort; K–M curves for 30-day (C) and 90-day (D) ACM in the HBUSH cohort Relationship between CAR and ACM To examine the impact of CAR on ACM, Cox regression analyses were performed (Tables 2, 3, 4 and 5, Supplementary Tables S4-S7). In the MIMIC-IV cohort, when CAR was analyzed as a continuous variable, all models demonstrated that every unit rise in CAR was linked to 3% elevated risks of 30-day and 90-day ACM (p ≤ 0.001). When CAR was analyzed as a categorical variable, individuals in Q3 (3.64 ≤ CAR < 6.35) had an approximately 145% increased risk of death compared with Q1 (CAR < 2.40) (hazard ratio [HR] = 2.45, 95% confidence interval [CI]: 1.74–3.45), while those in Q4 (CAR ≥ 6.35) had an approximately 211% elevated risk of death compared with Q1 (HR = 3.11, 95% CI: 2.23–4.35). After adjusting for age and sex (Model 2), the effect sizes slightly decreased but remained significant in Q3 and Q4 (Q3: HR = 2.33, 95% CI: 1.65–3.31; Q4: HR = 3.02, 95% CI: 2.16–4.23). After further adjustment for vital signs, comorbidities, medications, and laboratory indicators (Model 3), the risk of death remained independently elevated in Q3 (HR = 1.62, 95% CI: 1.13–2.32) and Q4 (HR = 2.17, 95% CI: 1.52–3.10). Similar trends were found for 90-day ACM and in the HMUSH cohort, further supporting the association between CAR and ACM. Table 2Cox proportional hazard models for 30-day all-cause mortality in MIMIC-IVVariablesModel1HR(95%CI) P Model2HR(95%CI) P Model3HR(95%CI) P CAR continuous1.03(1.02,1.04)< 0.0011.03(1.02,1.04)< 0.0011.03(1.01,1.04)0.001CAR quantileQ1(CAR<2.40)1.00(Reference)1.00(Reference)1.00(Reference)Q2(2.40 ≤ CAR<3.64)1.30(0.89,1.91)0.1731.26(0.86,1.85)0.2431.24(0.84,1.82)0.277Q3(3.64 ≤ CAR<6.35)2.45(1.74,3.45)< 0.0012.33(1.65,3.31)< 0.0011.62(1.13,2.32)0.009Q4(CAR ≥ 6.35)3.11(2.23,4.35)< 0.0013.02(2.16,4.23)< 0.0012.17(1.52,3.10)< 0.001HR for trend1.49(1.35,1.65)1.48(1.34,1.64)1.30(1.17,1.46)P for trend< 0.001< 0.001< 0.001HR Hazard Ratio, CI Confidence IntervalMolde1: No adjustedModel2: Age, GenderModel3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Table 2Cox proportional hazard models for 30-day all-cause mortality in MIMIC-IVVariablesModel1HR(95%CI) P Model2HR(95%CI) P Model3HR(95%CI) P CAR continuous1.03(1.02,1.04)< 0.0011.03(1.02,1.04)< 0.0011.03(1.01,1.04)0.001CAR quantileQ1(CAR<2.40)1.00(Reference)1.00(Reference)1.00(Reference)Q2(2.40 ≤ CAR<3.64)1.30(0.89,1.91)0.1731.26(0.86,1.85)0.2431.24(0.84,1.82)0.277Q3(3.64 ≤ CAR<6.35)2.45(1.74,3.45)< 0.0012.33(1.65,3.31)< 0.0011.62(1.13,2.32)0.009Q4(CAR ≥ 6.35)3.11(2.23,4.35)< 0.0013.02(2.16,4.23)< 0.0012.17(1.52,3.10)< 0.001HR for trend1.49(1.35,1.65)1.48(1.34,1.64)1.30(1.17,1.46)P for trend< 0.001< 0.001< 0.001HR Hazard Ratio, CI Confidence IntervalMolde1: No adjustedModel2: Age, GenderModel3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Cox proportional hazard models for 30-day all-cause mortality in MIMIC-IV HR Hazard Ratio, CI Confidence Interval Molde1: No adjusted Model2: Age, Gender Model3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Table 3Cox proportional hazard models for 90-day all-cause mortality in MIMIC-IVVariablesModel1HR(95%CI) P Model2HR(95%CI) P Model3HR(95%CI) P CAR continuous1.03(1.02,1.04)< 0.0011.03(1.02,1.04)< 0.0011.03(1.02,1.04)< 0.001CAR quantileQ1(CAR<2.40)1.00(Reference)1.00(Reference)1.00(Reference)Q2(2.40 ≤ CAR<3.64)1.36(0.99,1.88)0.0601.30(0.94,1.79)0.1171.32(0.95,1.83)0.094Q3(3.64 ≤ CAR<6.35)2.44(1.82,3.28)< 0.0012.28(1.69,3.08)< 0.0011.65(1.21,2.25)0.002Q4(CAR ≥ 6.35)3.05(2.29,4.06)< 0.0012.92(2.19,3.90)< 0.0012.18(1.61,2.95)< 0.001HR for trend1.47(1.35,1.60)1.46(1.34,1.59)1.29(1.18,1.42)P for trend< 0.001< 0.001< 0.001HR Hazard Ratio, CI Confidence IntervalMolde1: No adjustedModel2: Age, GenderModel3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Table 3Cox proportional hazard models for 90-day all-cause mortality in MIMIC-IVVariablesModel1HR(95%CI) P Model2HR(95%CI) P Model3HR(95%CI) P CAR continuous1.03(1.02,1.04)< 0.0011.03(1.02,1.04)< 0.0011.03(1.02,1.04)< 0.001CAR quantileQ1(CAR<2.40)1.00(Reference)1.00(Reference)1.00(Reference)Q2(2.40 ≤ CAR<3.64)1.36(0.99,1.88)0.0601.30(0.94,1.79)0.1171.32(0.95,1.83)0.094Q3(3.64 ≤ CAR<6.35)2.44(1.82,3.28)< 0.0012.28(1.69,3.08)< 0.0011.65(1.21,2.25)0.002Q4(CAR ≥ 6.35)3.05(2.29,4.06)< 0.0012.92(2.19,3.90)< 0.0012.18(1.61,2.95)< 0.001HR for trend1.47(1.35,1.60)1.46(1.34,1.59)1.29(1.18,1.42)P for trend< 0.001< 0.001< 0.001HR Hazard Ratio, CI Confidence IntervalMolde1: No adjustedModel2: Age, GenderModel3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Cox proportional hazard models for 90-day all-cause mortality in MIMIC-IV HR Hazard Ratio, CI Confidence Interval Molde1: No adjusted Model2: Age, Gender Model3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Table 4Cox proportional hazard models for 30-day all-cause mortality in HMUSHVariablesModel1HR(95%CI) P Model2HR(95%CI) P Model3HR(95%CI) P CAR continuous1.08(1.03,1.13)< 0.0011.09(1.04,1.14)< 0.0011.06(1.00,1.11)0.039CAR quantileQ1(CAR<1.950)1.00(Reference)1.00(Reference)1.00(Reference)Q2(1.950 ≤ CAR<2.585)1.53(0.63,3.75)0.3511.58(0.63,3.93)0.3281.93(0.77,4.80)0.159Q3(2.585 ≤ CAR<3.930)1.68(0.70,4.06)0.2461.68(0.66,4.26)0.2741.68(0.64,4.36)0.290Q4(CAR ≥ 3.930)3.41(1.54,7.57)0.0033.43(1.49,7.89)0.0042.64(1.09,6.41)0.032HR for trend1.50(1.17,1.92)1.50(1.17,1.93)1.32(1.01,1.74)P for trend0.0010.0020.043HR: Hazard Ratio, CI: Confidence IntervalMolde1: No adjustedModel2: Age, GenderModel3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Table 4Cox proportional hazard models for 30-day all-cause mortality in HMUSHVariablesModel1HR(95%CI) P Model2HR(95%CI) P Model3HR(95%CI) P CAR continuous1.08(1.03,1.13)< 0.0011.09(1.04,1.14)< 0.0011.06(1.00,1.11)0.039CAR quantileQ1(CAR<1.950)1.00(Reference)1.00(Reference)1.00(Reference)Q2(1.950 ≤ CAR<2.585)1.53(0.63,3.75)0.3511.58(0.63,3.93)0.3281.93(0.77,4.80)0.159Q3(2.585 ≤ CAR<3.930)1.68(0.70,4.06)0.2461.68(0.66,4.26)0.2741.68(0.64,4.36)0.290Q4(CAR ≥ 3.930)3.41(1.54,7.57)0.0033.43(1.49,7.89)0.0042.64(1.09,6.41)0.032HR for trend1.50(1.17,1.92)1.50(1.17,1.93)1.32(1.01,1.74)P for trend0.0010.0020.043HR: Hazard Ratio, CI: Confidence IntervalMolde1: No adjustedModel2: Age, GenderModel3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Cox proportional hazard models for 30-day all-cause mortality in HMUSH HR: Hazard Ratio, CI: Confidence Interval Molde1: No adjusted Model2: Age, Gender Model3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Table 5Cox proportional hazard models for 90-day all-cause mortality in HMUSHVariablesModel1HR(95%CI) P Model2HR(95%CI) P Model3HR(95%CI) P CAR continuous1.07(1.03,1.12)< 0.0011.08(1.04,1.13)< 0.0011.05(1.00,1.10)0.040CAR quantileQ1(CAR<1.950)1.00(Reference)1.00(Reference)1.00(Reference)Q2(1.950 ≤ CAR<2.585)1.25(0.61,2.53)0.5421.32(0.64,2.72)0.4561.65(0.80,3.40)0.173Q3(2.585 ≤ CAR<3.930)1.25(0.62,2.54)0.5331.28(0.60,2.70)0.5221.38(0.64,2.98)0.415Q4(CAR ≥ 3.930)2.82(1.52,5.25)0.0012.89(1.50,5.56)0.0022.39(1.18,4.83)0.015HR for trend1.42(1.16,1.74)1.44(1.17,1.77)1.30(1.04,1.63)P for trend< 0.001< 0.0010.022HR Hazard Ratio, CI Confidence IntervalMolde1: No adjustedModel2: Age, GenderModel3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Table 5Cox proportional hazard models for 90-day all-cause mortality in HMUSHVariablesModel1HR(95%CI) P Model2HR(95%CI) P Model3HR(95%CI) P CAR continuous1.07(1.03,1.12)< 0.0011.08(1.04,1.13)< 0.0011.05(1.00,1.10)0.040CAR quantileQ1(CAR<1.950)1.00(Reference)1.00(Reference)1.00(Reference)Q2(1.950 ≤ CAR<2.585)1.25(0.61,2.53)0.5421.32(0.64,2.72)0.4561.65(0.80,3.40)0.173Q3(2.585 ≤ CAR<3.930)1.25(0.62,2.54)0.5331.28(0.60,2.70)0.5221.38(0.64,2.98)0.415Q4(CAR ≥ 3.930)2.82(1.52,5.25)0.0012.89(1.50,5.56)0.0022.39(1.18,4.83)0.015HR for trend1.42(1.16,1.74)1.44(1.17,1.77)1.30(1.04,1.63)P for trend< 0.001< 0.0010.022HR Hazard Ratio, CI Confidence IntervalMolde1: No adjustedModel2: Age, GenderModel3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT Cox proportional hazard models for 90-day all-cause mortality in HMUSH HR Hazard Ratio, CI Confidence Interval Molde1: No adjusted Model2: Age, Gender Model3: Age, Gender, HR, SBP, DBP, RR, Diabetes Mellitus, Hypertension, Pneumonia, Liver Cirrhosis, Ascites, Antiplatelet Drugs, Albumin Infusion, Antimicrobial Drugs, AST, INR, HB, PLT RCS analysis RCS analysis demonstrated a dose-response association between CAR and 30-day and 90-day ACM (Fig. 3). In the MIMIC-IV cohort, the association appeared nonlinear. In contrast, the association appeared linear in the HMUSH cohort. Fig. 3RCS analysis of the relationship between CAR and ACM. RCS curves for 30-day (A) and 90-day (B) ACM in the MIMIC-IV cohort; RCS curves for 30-day (C) and 90-day (D) ACM in the HBUSH cohort RCS analysis of the relationship between CAR and ACM. RCS curves for 30-day (A) and 90-day (B) ACM in the MIMIC-IV cohort; RCS curves for 30-day (C) and 90-day (D) ACM in the HBUSH cohort Subgroup analyses To examine whether the association between CAR values and 30-day and 90-day ACM persisted across different subgroups, subgroup analyses based on age (≥ 65 or < 65 years), sex (female or male), DM, HTN, cirrhosis, pulmonary infection, ascites, use of antiplatelet drugs, ALB infusion, and antibiotic use were carried out (Fig. 4). Fig. 4Forest plots illustrating subgroup analyses of the association between CAR and ACM. Subgroup analyses for 30-day (A) and 90-day (B) ACM in the MIMIC-IV cohort; Subgroup analyses for 30-day (C) and 90-day (D) ACM in the HBUSH cohort Forest plots illustrating subgroup analyses of the association between CAR and ACM. Subgroup analyses for 30-day (A) and 90-day (B) ACM in the MIMIC-IV cohort; Subgroup analyses for 30-day (C) and 90-day (D) ACM in the HBUSH cohort In the MIMIC-IV cohort, the associations between CAR and 30-day and 90-day ACM among UGIB patients remained significant across nearly all subgroups, except among females. In the HMUSH cohort, a significantly elevated risk of 30-day ACM was found among UGIB patients with HTN, pulmonary infection, those without long-term administration of antiplatelet drugs, those receiving ALB infusion and antibiotics during ICU stays. Elevated risk of 90-day ACM was found among patients aged ≥ 65, female patients, those with pulmonary infection, ascites, those without long-term administration of antiplatelet drugs, and those receiving ALB infusion and antibiotics during ICU stays. Interaction analyses demonstrated that sex, age, DM, HTN, pulmonary infection, ascites, use of antiplatelet drugs, ALB infusion, and antibiotic use did not markedly modify the relationship between CAR and ACM (P for interaction > 0.05). Notably, concomitant cirrhosis significantly influenced this relationship, with interactions found for both 30-day and 90-day ACM (P for interaction < 0.05). Discussion This study, based on the MIMIC-IV and HMUSH cohorts, demonstrated an independent relationship between CAR and 30-day and 90-day ACM in individuals with UGIB. Subgroup and interaction analyses demonstrated that these findings were robust. These results could help identify patients at high risk and improve clinical outcomes. Cr is an easily accessible serum marker reflecting renal function. Prior studies have demonstrated that elevated Cr levels predict poor prognosis in UGIB patients [17, 18]. During UGIB, particularly in cases of massive bleeding, patients experience substantial blood volume loss. This results in decreased renal perfusion pressure, ischemia of the renal tubules and interstitium, and subsequent direct renal tissue injury [17]. In addition, following UGIB, the autonomic nervous system is activated, such as the antidiuretic hormone, the renin–angiotensin–aldosterone system, and the sympathetic nervous system. This condition induces renal vasoconstriction, further decreases renal blood flow, aggravates ischemia, and ultimately leads to the development of AKI [23, 24]. UGIB is also frequently accompanied by complications such as bacterial infection. Inflammatory response damages the kidney through cytokines and oxidative stress, leading to tubular necrosis or tubulointerstitial fibrosis [25, 26]. Moreover, the inflammatory response increases systemic microvascular permeability, resulting in impaired microcirculatory blood flow and further exacerbating renal ischemia and injury [27, 28]. These findings suggest that once tubular function is impaired, the kidney fails to effectively eliminate metabolic waste from the blood, including Cr, thereby causing its serum concentration to rise. ALB is closely linked to oxidative stress and nutritional status [29]. UGIB is often accompanied by systemic inflammation, which activates the acute-phase response and promotes the synthesis of acute-phase proteins (including fibrinogen and C-reactive protein). In contrast, ALB, as a negative acute-phase protein, shows reduced synthesis and increased consumption [10, 29]. Inflammatory response also increases vascular permeability, leading to extravasation of ALB [30]. Clinically, UGIB patients frequently receive large-volume fluid resuscitation, which leads to hemodilution and a subsequent decline in plasma ALB concentration [31]. Moreover, during the acute bleeding phase, UGIB patients are often fasted, which results in a deficiency of essential amino acids. This deficiency directly impairs hepatic ALB synthesis and consequently leads to a decline in plasma ALB levels [29]. CAR not only reflects renal dysfunction and malnutrition, but also systemic inflammation and redox balance. Renal impairment reduces clearance of inflammatory mediators, leading to a marked increase in inflammatory factors, including tumor necrosis factor-α and interleukin-6 [32], which disrupt vascular endothelial integrity and gut barrier function. Hypoalbuminemia, by inhibiting the activity of macrophages, B cells, and T cells, limits immune cell proliferation, differentiation, and function, thereby weakening both the strength and duration of immune responses [33]. The inflammatory response further exacerbates renal injury and ALB loss, creating a vicious cycle that worsens disease severity. The roles of creatinine and albumin in oxidative stress have gained increasing attention. Recent studies have found that under oxidative stress conditions, such as oxidative damage caused by smoking, changes in creatinine levels may indirectly reflect alterations in redox balance [34, 35]. ALB is an important endogenous antioxidant. By binding metal ions, it inhibits the generation of reactive oxygen species and directly scavenges free radicals, exerting anti-inflammatory effects and playing a crucial role in antioxidant defense [36]. These findings support the potential role of CAR in oxidative stress. CAR has several advantages: First, Cr and ALB are simple and inexpensive to measure, making CAR widely applicable in hospitalized patients. In addition, in ICU patients with UGIB, increased CAR values usually suggest an elevated risk of complications. This underscores the importance of timely optimization of fluid management to avoid both over- and underhydration, as well as active nutritional support, including early parenteral or enteral nutrition and targeted supplementation of specific nutrients to improve protein metabolism. In this study, concomitant cirrhosis modified the association between the CAR and 30-day and 90-day ACM. The progression of hepatic fibrosis or cirrhosis obstructs hepatic blood flow, leading to increased portal venous pressure and subsequent visceral vasodilation, particularly in the splanchnic circulation. Despite this vasodilation, the total circulating blood volume does not increase; instead, systemic blood redistribution occurs, with blood shifting from organs such as the kidneys to the abdominal viscera, leading to renal hypoperfusion [37]. In addition, patients with cirrhosis frequently present with adrenal cortical insufficiency [38]. Cortisol deficiency reduces vascular reactivity, impairs blood pressure regulation, and induces renal vasodilation with insufficient perfusion. Aldosterone deficiency causes fluid and electrolyte imbalance, hypovolemia, and hypotension, thereby directly reducing renal blood flow [39, 40]. ALB is synthesized in the liver [41], and impaired hepatic function reduces its synthetic capacity. Moreover, cirrhotic patients often have inadequate dietary intake due to anorexia, dyspepsia, or gastrointestinal edema. Additionally, cirrhotic patients may also develop hepatorenal syndrome, which diminishes renal filtration capacity. In the presence of tubular injury or tubulointerstitial edema, ALB and other proteins may be lost via the urine [42]. The relationship between CAR and 30-day or 90-day ACM was not consistent between the MIMIC-IV and HBUSH cohorts. Several factors may account for this discrepancy: (i) baseline differences in race, comorbidities, and other clinical characteristics between cohorts; (ii) heterogeneity in the etiologies of UGIB between cohorts, which were not investigated through subgroup analyses in this study; (iii) differences in disease severity between the two cohorts; and (iv) variations in therapeutic strategies, such as the timing of endoscopic intervention. This dual-center study reveals that there is an association between CAR and the prognosis of UGIB patients, thereby facilitating early detection of patients at high risk and guiding individualized treatment to improve prognosis. However, some limitations must be noted. First, as we mentioned above, there is heterogeneity between the two cohorts, which could affect the generalizability of our findings across different populations. Second, HBUSH lacked disease severity scores, which restricted pathological assessment. Thirdly, the HMUSH cohort has a relatively small sample size, which may increase the risk of overfitting in the multivariable models. These results need to be further validated in larger, independent cohorts to confirm their robustness and generalizability. Additionally, some important confounders, such as lifestyle factors, nutritional status, and inflammatory indices, were not included in our analysis. Hence, there are potential residual confounders due to unmeasured variables. Furthermore, because the dataset primarily represents ICU patients, the findings may not be directly generalizable to the broader population, and caution is warranted when extrapolating these results. The substantial missing data for inflammatory markers, blood lipids, and blood glucose also limited our investigation of the underlying pathological pathways. Finally, as this study is retrospective, causal inference is inherently limited. Future prospective research is warranted to confirm these findings. Conclusion This research suggests that CAR is a high-risk factor for 30-day and 90-day ACM among individuals with UGIB. The findings provide a close association between serum indicators and the prognosis of UGIB patients, but further studies are required to confirm this result and to investigate how CAR-based strategies could improve patient outcomes. Supplementary Information Supplementary Material 1: Table S1. The number of missing data in the MIMIC-IV and HMUSH cohorts. Table S2. Variance Inflation Factor Values for the Covariates of the Relationship between 30-day All-cause Mortality of UGIB patients and CAR Group with PT included.Table S3. Variance Inflation Factor Values for the Covariates of the Relationship between 30-day All-cause Mortality of UGIB patients and CAR Group with PT excluded.Table S4. Cox analysis of the factors influencing 30-day all-cause mortality in MIMIC-IV.Table S5. Cox analysis of the factors influencing 90-day all-cause mortality in MIMIC-IV.Table S6. Cox analysis of the factors influencing 30-day all-cause mortality in HMUSH.Table S7. Cox analysis of the factors influencing 90-day all-cause mortality in HMUSH Supplementary Material 1: Table S1. The number of missing data in the MIMIC-IV and HMUSH cohorts. Table S2. Variance Inflation Factor Values for the Covariates of the Relationship between 30-day All-cause Mortality of UGIB patients and CAR Group with PT included.Table S3. Variance Inflation Factor Values for the Covariates of the Relationship between 30-day All-cause Mortality of UGIB patients and CAR Group with PT excluded.Table S4. Cox analysis of the factors influencing 30-day all-cause mortality in MIMIC-IV.Table S5. Cox analysis of the factors influencing 90-day all-cause mortality in MIMIC-IV.Table S6. Cox analysis of the factors influencing 30-day all-cause mortality in HMUSH.Table S7. Cox analysis of the factors influencing 90-day all-cause mortality in HMUSH