Systolic blood pressure time in target range within 24 h and in-hospital death of critically ill patients and the role of ventricular arrhythmias: a MIMIC-IV analysis with external validation in the eICU-CRD database Authors: Qiwei Shen, Hailong Li, Peng Liu, Dianfei Long, Lei Wang, Yilong Man Published: 2025 ABSTRACT BackgroundThe stability of blood pressure control is crucial in critically ill patients. Systolic blood pressure (SBP) time in target range (TTR) has been recognized as an emerging dynamic assessment indicator of blood pressure, but its prognostic value in critically ill patients remains unclear.MethodsThis retrospective cohort study utilized data from the MIMIC-IV database, including adult patients with their first intensive care unit (ICU) admission. SBP TTR within 110–140 mmHg during the first 24 h was calculated. The primary outcome was in-hospital all-cause mortality, and the secondary outcome was ventricular arrhythmias (VA, including FPVCs, VT, and VF). Logistic regression, restricted cubic splines, subgroup analyses, receiver operating characteristic (ROC) curve analyses, and Karlson–Holm–Breen mediation analysis were performed. Furthermore, the findings were externally validated using the multicenter eICU-CRD database.ResultsA total of 42,648 patients were included in the primary analysis. Higher SBP TTR was significantly associated with lower in-hospital all-cause mortality (adjusted OR 0.94 per 10% increase, 95% CI 0.93–0.95) and reduced risk of VA (adjusted OR 0.97 per 10% increase, 95% CI 0.95–0.98). When compared with the lowest quartile, patients in the highest SBP TTR quartile demonstrated the lowest risk of in-hospital all-cause mortality (OR 0.66) and VA (OR 0.78). Mediation analysis suggested that VA mediated a small proportion (1.96%) of the effect of SBP TTR on in-hospital all-cause mortality. Furthermore, SBP TTR provided prognostic information independent of SBP coefficient of variation, and combining both metrics improved predictive accuracy (AUC = 0.62). Finally, the association between SBP TTR and the primary outcome (in-hospital all-cause mortality) was successfully validated in the multicenter eICU-CRD cohort (n = 55,155).ConclusionsHigher SBP TTR (110–140 mmHg) during the first 24 h of ICU admission was independently and inversely associated with in-hospital mortality and ventricular arrhythmias. This metric provides information complementary to blood pressure variability. The association with in-hospital mortality was consistently validated in the multicenter eICU-CRD database, suggesting SBP TTR may serve as a useful biomarker for risk stratification in ICU patients.Graphical Abstract Supplementary InformationThe online version contains supplementary material available at 10.1186/s40001-025-03691-0. Introduction Hemodynamic instability is common in critically ill patients, and blood pressure often shows marked fluctuations during disease progression and treatment [1–3]. Such variability can compromise organ perfusion and oxygen delivery, and may also trigger serious adverse cardiovascular events [4, 5]. In the intensive care unit (ICU) setting, malignant ventricular arrhythmias (VA), including frequent premature ventricular contractions (FPVCs), ventricular tachycardia (VT), and ventricular fibrillation (VF), occur frequently and are strongly associated with in-hospital mortality [6, 7]. Previous studies suggest that blood pressure variability may be an important trigger for VA [8, 9]. However, conventional assessments of blood pressure largely rely on single values or average levels. These measures do not adequately capture the “stability” of blood pressure control and have limited predictive value for arrhythmias and outcomes. Time in target range (TTR) for systolic blood pressure (SBP) has recently emerged as a dynamic metric [10, 11]. It reflects the proportion of time blood pressure remains within the desired range, thereby providing a more accurate representation of blood pressure stability. In chronic disease management, such as diabetes and hypertension, TTR has been shown to be closely linked to clinical outcomes [12]. Yet, its value in critically ill patients remains unclear, particularly regarding its association with VA and in-hospital mortality. To address this gap, we utilized the large-scale MIMIC-IV clinical database to analyze patients during their first ICU admission. We aimed to evaluate the association between SBP TTR within the first 24 h of admission and in-hospital all-cause mortality, as well as VA. Furthermore, we compared the prognostic value of SBP TTR with the Systolic Blood Pressure Coefficient of Variation (SBP CV) to assess their complementarity, and explored whether VA plays a mediating role in the relationship between SBP TTR and mortality. To ensure the robustness of our findings, an external validation was performed using the multicenter eICU-CRD database. These findings could provide new insights into dynamic blood pressure assessment and risk stratification in critically ill patients. Methods Study design and data source This retrospective study was conducted using data from the Medical Information Mart for Intensive Care IV (MIMIC-IV, version 3.1) database (https://mimic.mit.edu) [13]. MIMIC-IV contains comprehensive clinical data for patients admitted to the emergency department and ICU of Beth Israel Deaconess Medical Center (BIDMC) between 2008 and 2022. The database includes records from 364,627 patients, of whom 223,452 had a first hospital admission documented, and 65,366 represented initial ICU admissions. Data extraction and study population Only patients with their first ICU admission were included in this analysis. Detailed inclusion and exclusion criteria are provided in Fig. 1. Data extraction was performed by an approved researcher, Qiwei Shen (Certification Number: 59,548,768). The SQL code used for data querying and extraction is publicly available through the MIMIC Code Repository. Demographic information (age, sex, and race), body mass index (BMI), ICU stay time, Charlson comorbidity index, vital signs (systolic pressure, diastolic pressure, mean arterial pressure, heart rate), laboratory values (potassium, sodium, total calcium, blood urea nitrogen (BUN), creatinine, aspartate aminotransferase (AST), alanine aminotransferase (ALT), international normalized ratio (INR), glucose, triglyceride (TG), total cholesterol (TC), high-density lipoprotein (HDL), low-density lipoprotein (LDL), lactate, albumin, C-reactive protein (CRP), white blood cell count (WBC), red blood cell count (RBC), platelet, hemoglobin), comorbidities (hypertension, diabetes, heart failure, coronary artery disease, atrial fibrillation, cerebrovascular disease, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, sedative medications, mechanical ventilation, mechanical circulatory support) were extracted using Structured Query Language (SQL) and Navicat Premium version 15.0. Comorbidities were identified through ICD-9 or ICD-10 discharge diagnosis codes. Vital signs and laboratory values were both extracted from the first recorded measurements after patient admission.Fig. 1Flow chart of patient selection Flow chart of patient selection Missing data were addressed using multiple imputation with the missForest package in R. For variables with more than 25% missingness, dummy coding was applied in the regression models instead of direct imputation to minimize potential bias. Variables with missingness greater than 25% included BMI, albumin, CRP, lactate, AST, ALT, HDL-C, LDL-C, TG, and TC. Exposure and outcomes The exposure of interest was the SBP TTR during the first 24 h after the initial ICU admission. SBP values were obtained from the vital sign table, and both invasive arterial blood pressure measurements and noninvasive cuff blood pressure measurements were included in the analysis. In the ICU, blood pressure is typically recorded every 30 min to 1 h, providing sufficient temporal resolution for time-based assessments. The SBP target range was defined as 110–140 mmHg [14, 15]. TTR was calculated using linear interpolation across all available SBP measurements within the first 24 h window [16, 17]. To ensure sufficient data for reliable estimation, patients were included in the TTR analysis only if SBP recordings were available for at least 19 h within this 24 h window. The primary outcome was in-hospital all-cause mortality. The secondary outcome was VA. VA defined as FPVCs, VT, and VF, will be identified using the rhythm table. Mortality will be identified using the date of death recorded in the patients table. To ensure a clear temporal relationship between exposure and outcomes, patients with an ICU length of stay less than 24 h were excluded from the analysis. Statistical analyses For normally distributed continuous variables, data are presented as mean ± standard deviation (Mean ± SD) and compared across groups using one-way ANOVA. For non-normally distributed continuous variables, data are summarized as median (interquartile range), and group differences are assessed with the Kruskal–Wallis test. Categorical variables are presented as frequencies and percentages, with differences evaluated by the chi-square test. Collinearity among all baseline variables was first assessed, and variables with high collinearity were selectively excluded. Variables with a variance inflation factor (VIF) > 5 or a tolerance < 0.2 were considered highly collinear and removed from the analysis. Specifically, DBP, TC, HDL, AST, and Hemoglobin were excluded. The results of collinearity assessment for all baseline variables are shown in Supplementary Table 1, and the collinearity analysis for the covariates included in the final models is presented in Supplementary Table 2. The association between SBP TTR and clinical outcomes was assessed using logistic regression models for in-hospital all-cause mortality (primary outcome) and VA (secondary outcome). SBP TTR was analyzed both as a continuous variable (per 10% increase) and as a categorical variable based on quartiles. Three models were constructed: Model I was unadjusted; Model II was adjusted for age, sex, and BMI; and Model III was further adjusted for race, ICU stay time, Charlson comorbidity index, vital signs (heart rate, systolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, ALT, albumin, TG, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, sedative medications, mechanical ventilation, mechanical circulatory support). We further constructed a model adjusting for SBP coefficient of variation (SBP CV) to assess whether SBP TTR remained independently associated with outcomes after accounting for blood pressure variability. To evaluate the discriminative performance of SBP TTR and SBP CV, receiver operating characteristic (ROC) curves were generated for five models: SBP alone, MBP alone, SBP CV alone, SBP TTR alone, and SBP TTR combined with SBP CV. The area under the ROC curve (AUC) was compared among these models to determine whether SBP TTR improved predictive accuracy independently of SBP CV. To examine whether the relationship between SBP TTR and in-hospital all-cause mortality is mediated by VA, we used the KHB (Karlson–Holm–Breen) method, which decomposes the total effect of SBP TTR into direct and VA-mediated indirect effects, while adjusting for confounders [18]. Subgroup analyses were conducted by age (< 65 years), gender, race, hypertension, diabetes, and sepsis. Restricted cubic splines (RCS) were used to assess potential nonlinear associations between SBP TTR and outcomes, with four knots placed at the 5th, 35th, 65th, and 95th percentiles of the SBP TTR distribution. Additionally, three sensitivity analyses were performed. First, SBP TTR was calculated using the proportion method, and its association with clinical outcomes was evaluated. Second, an analysis was conducted after excluding covariates with more than 25% missing data [17]. Third, to assess the robustness of the secondary outcome definition, FPVCs and VT/VF were analyzed separately as individual outcomes. Finally, the association between SBP TTR and the primary outcome was independently validated using the eICU-CRD database. All analyses were performed with R version 4.4.1, except for the KHB mediation analysis, which was conducted in Stata 17.0. Statistical significance was set at P < 0.05. Externally validated in eICU-CRD An independent external validation was performed using the eICU-CRD as a separate cohort. The database, which includes clinical records from 208 ICUs across the United States between 2014 and 2015, contains extensive biochemical and physiological monitoring data [19]. The same inclusion and exclusion criteria, variable definitions, outcome measures, and statistical methods as those applied in the MIMIC-IV analysis were adopted. However, because the eICU-CRD does not contain rhythm data from continuous ECG monitoring, the external validation was restricted to the primary outcome. Through this independent cohort analysis, the robustness, generalizability, and consistency of the findings were further evaluated. Results A total of 42,648 ICU patients were included in the analysis. The median age was 66.6 years, 43.9% were female, and 65.0% were White. The prevalence of comorbidities included hypertension in 63.8%, diabetes in 28.6%, and sepsis in 16.8%. Overall, 10.5% of patients experienced in-hospital all-cause mortality, and 9.0% developed VA. Patients in the highest quartile of SBP TTR (Q4) had the lowest incidence of both in-hospital mortality and VA (Fig. 2A, B). Detailed baseline characteristics stratified by SBP TTR quartiles are summarized in Table 1.Fig. 2Association between SBP TTR and in-hospital all-cause mortality. A Incidence of in-hospital mortality across quartiles of SBP TTR. B Incidence of VA across quartiles of SBP TTR. C Subgroup analyses of the association between SBP TTR (per 10% increase) and in-hospital mortality. ORs with 95% CIs were estimated using multivariable logistic regression. D RCS curves showing the inverse association between SBP TTR and in-hospital mortality. P values for overall and nonlinearity are shownTable 1Baseline characteristics of ICU patients according to quartile of SBP TTRCharacteristicsTotal (n = 42,648)Q1, (n = 9,969)Q2, (n = 11,981)Q3, (n = 13,158)Q4, (n = 7,540)P valueAge, years66.6 (54.8–77.5)64.9 (53.7–76.1)67.4 (56.7–78.2)67.9 (56.1–78.7)64.4 (51.0–75.7) < 0.001Sex, female, n (%)18706 (43.9)4515 (45.3)5393 (45.0)5739 (43.6)3059 (40.6) < 0.001BMI, kg/m227.7 (24.1–32.2)27.0 (23.4–31.6)27.9 (24.2–32.4)27.9 (24.3–32.6)28.0 (24.4–32.3) < 0.001Charlson comorbidity index5.0 (3.0–7.0)5.0 (3.0–7.0)5.0 (3.0–7.0)5.0 (3.0–7.0)4.0 (2.0–6.0) < 0.001ICU stay time, days2.3 (1.5–4.4)2.6 (1.7–4.9)2.5 (1.6–4.6)2.3 (1.5–4.2)2.1 (1.4–3.8) < 0.001SBP CV11.3 (9.0–14.0)10.5 (8.4–13.4)12.9 (10.3–16.0)12.0 (10.0–14.2)9.0 (7.5–10.6) < 0.001Race, n (%) White27,700 (65.0)6501 (65.2)7776 (64.9)8550 (65.0)4873 (64.6)0.883 Other/unknown14,948 (35.0)3468 (34.8)4205 (35.1)4608 (35.0)2667 (35.4)Vital signs Heart rate, beats/min86.0 (75.0–100.0)89.0 (77.0–104.0)85.0 (74.0–100.0)84.0 (74.0–98.0)84.0 (73.0–98.0) < 0.001 Systolic BP, mmHg123.0 (108.0–140.0)110.0 (98.0–139.0)121.0 (106.0–142.0)126.0 (112.0–142.0)128.0 (118.0–138.0) < 0.001 Diastolic BP, mmHg67.0 (57.0–79.0)65.0 (55.0–78.0)67.0 (56.0–79.0)67.0 (57.0–79.0)69.0 (60.0–80.0) < 0.001 Mean arterial pressure, mmHg83.0 (72.0–96.0)79.0 (68.0–94.0)83.0 (72.0–96.0)84.0 (74.0–96.0)86.0 (77.0–95.0) < 0.001Laboratory examinations Potassium, mEq/L4.1 (3.8–4.5)4.1 (3.7–4.6)4.1 (3.8–4.5)4.1 (3.8–4.5)4.1 (3.8–4.5) < 0.001 Sodium, mEq/L139.0 (136.0–141.0)138.0 (135.0–141.0)139.0 (136.0–141.0)139.0 (136.0–141.0)139.0 (136.0–141.0) < 0.001 Total calcium, mg/dL8.5 (8.0–9.0)8.4 (7.8–8.9)8.4 (7.9–8.9)8.5 (8.0–9.0)8.6 (8.1–9.0) < 0.001 Creatinine, mg/dL0.9 (0.7–1.3)1.0 (0.7–1.5)1.0 (0.7–1.4)0.9 (0.7–1.3)0.9 (0.7–1.2) < 0.001 BUN, mg/dL18.0 (13.0–28.0)19.0 (13.0–32.0)18.0 (13.0–29.0)18.0 (13.0–27.0)16.0 (12.0–24.0) < 0.001 AST, IU/L33.0 (21.0–67.0)36.0 (22.0–81.0)34.0 (21.0–70.0)31.0 (21.0–60.0)31.0 (20.0–61.0) < 0.001 ALT, IU/L25.0 (15.0–48.0)26.0 (15.0–53.0)25.0 (15.0–50.0)24.0 (15.0–44.5)25.0 (16.0–45.0) < 0.001 INR1.2 (1.1–1.4)1.2 (1.1–1.5)1.2 (1.1–1.4)1.2 (1.1–1.4)1.2 (1.1–1.4) < 0.001 Glucose, mg/dL124.0 (103.0–157.0)121.0 (100.0–155.0)125.0 (103.0–159.0)125.0 (104.0–159.0)124.0 (104.0–154.0) < 0.001 TG, mg/dL123.0 (85.0–190.0)119.0 (83.0–186.0)128.0 (86.0–198.0)124.0 (86.0–189.0)120.0 (84.0–185.0)0.014 TC, mg/dL148.0 (114.0–183.0)141.0 (105.0–176.0)147.0 (115.0–182.0)151.0 (118.0–185.0)153.0 (119.2–186.0) < 0.001 HDL, mg/dL42.0 (32.0–53.0)40.0 (30.0–52.0)42.0 (31.0–54.0)43.0 (33.0–54.0)42.0 (32.0–54.0) < 0.001 LDL, mg/dL78.0 (54.0–107.0)75.0 (49.0–105.0)76.0 (54.0–106.0)80.0 (56.0–109.0)80.0 (56.0–109.0)0.004 Lactate, mmol/L1.5 (1.1–2.3)1.7 (1.2–2.6)1.6 (1.1–2.3)1.5 (1.1–2.1)1.4 (1.1–2.1) < 0.001 Albumin, g/dL3.3 (2.8–3.8)3.1 (2.6–3.6)3.2 (2.8–3.7)3.4 (2.9–3.8)3.4 (3.0–3.9) < 0.001 CRP, mg/L72.3 (18.8–149.2)71.7 (24.5–149.7)76.3 (25.4–152.9)71.5 (15.9–154.6)59.8 (12.1–137.2)0.004 WBC, K/uL10.0 (7.3–14.0)10.2 (7.2–14.4)10.1 (7.2–14.2)10.0 (7.3–13.8)9.9 (7.3–13.5)0.016 RBC, K/uL3.8 (3.2–4.3)3.7 (3.1–4.2)3.7 (3.2–4.3)3.8 (3.3–4.3)3.9 (3.3–4.4) < 0.001 Platelet, K/uL200.0 (148.0–261.0)198.0 (142.0–263.0)199.0 (146.0–261.0)201.0 (151.0–262.0)201.0 (151.0–260.0)0.001 Hemoglobin, g/dL11.3 (9.6–12.9)11.0 (9.4–12.7)11.2 (9.6–12.8)11.4 (9.7–13.0)11.5 (9.8–13.1) < 0.001Comorbidities Coronary artery disease, n (%)14,470 (33.9)3180 (31.9)4366 (36.4)4656 (35.4)2268 (30.1) < 0.001 Atrial fibrillation, n (%)11,409 (26.8)2764 (27.7)3411 (28.5)3559 (27.0)1675 (22.2) < 0.001 Hypertension, n (%)27,216 (63.8)5755 (57.7)7922 (66.1)8913 (67.7)4626 (61.4) < 0.001 Heart failure, n (%)9713 (22.8)2786 (27.9)2757 (23.0)2812 (21.4)1358 (18.0) < 0.001 Diabetes mellitus, n (%)12,204 (28.6)2666 (26.7)3554 (29.7)3969 (30.2)2015 (26.7) < 0.001 Cerebrovascular disease, n (%)7266 (17.0)1328 (13.3)1883 (15.7)2546 (19.3)1509 (20.0) < 0.001 Chronic pulmonary disease, n (%)9931 (23.3)2374 (23.8)2887 (24.1)3083 (23.4)1587 (21.0) < 0.001 Renal disease, n (%)7702 (18.1)2008 (20.1)2263 (18.9)2310 (17.6)1121 (14.9) < 0.001 Tumors, n (%)6331 (14.8)1764 (17.7)1758 (14.7)1776 (13.5)1033 (13.7) < 0.001 Sepsis, n (%)7171 (16.8)2485 (24.9)2262 (18.9)1735 (13.2)689 (9.1) < 0.001Treatments Vasoactive agents, n (%)12,841 (30.1)3659 (36.7)4445 (37.1)3545 (26.9)1192 (15.8) < 0.001 Mechanical circulatory support, n (%)591 (1.4)219 (2.2)166 (1.4)124 (0.9)82 (1.1) < 0.001Outcomes In-hospital mortality, n (%)4481 (10.5)1465 (14.7)1335 (11.1)1178 (9.0)503 (6.7) < 0.001 Ventricular arrhythmias, n (%)3859 (9.0)1092 (11.0)1125 (9.4)1148 (8.7)494 (6.6) < 0.001ALT: alanine aminotransferase; AST: aspartate aminotransferase; BMI: body mass index; BP: blood pressure; BUN: blood urea nitrogen; CRP: C-reactive protein; HDL: high-density lipoprotein; ICU: intensive care unit; INR: international normalized ratio; LDL: low-density lipoprotein; RBC: red blood cell count; TC: total cholesterol; TG: triglyceride; WBC: white blood cell count Association between SBP TTR and in-hospital all-cause mortality. A Incidence of in-hospital mortality across quartiles of SBP TTR. B Incidence of VA across quartiles of SBP TTR. C Subgroup analyses of the association between SBP TTR (per 10% increase) and in-hospital mortality. ORs with 95% CIs were estimated using multivariable logistic regression. D RCS curves showing the inverse association between SBP TTR and in-hospital mortality. P values for overall and nonlinearity are shown Table 1Baseline characteristics of ICU patients according to quartile of SBP TTRCharacteristicsTotal (n = 42,648)Q1, (n = 9,969)Q2, (n = 11,981)Q3, (n = 13,158)Q4, (n = 7,540)P valueAge, years66.6 (54.8–77.5)64.9 (53.7–76.1)67.4 (56.7–78.2)67.9 (56.1–78.7)64.4 (51.0–75.7) < 0.001Sex, female, n (%)18706 (43.9)4515 (45.3)5393 (45.0)5739 (43.6)3059 (40.6) < 0.001BMI, kg/m227.7 (24.1–32.2)27.0 (23.4–31.6)27.9 (24.2–32.4)27.9 (24.3–32.6)28.0 (24.4–32.3) < 0.001Charlson comorbidity index5.0 (3.0–7.0)5.0 (3.0–7.0)5.0 (3.0–7.0)5.0 (3.0–7.0)4.0 (2.0–6.0) < 0.001ICU stay time, days2.3 (1.5–4.4)2.6 (1.7–4.9)2.5 (1.6–4.6)2.3 (1.5–4.2)2.1 (1.4–3.8) < 0.001SBP CV11.3 (9.0–14.0)10.5 (8.4–13.4)12.9 (10.3–16.0)12.0 (10.0–14.2)9.0 (7.5–10.6) < 0.001Race, n (%) White27,700 (65.0)6501 (65.2)7776 (64.9)8550 (65.0)4873 (64.6)0.883 Other/unknown14,948 (35.0)3468 (34.8)4205 (35.1)4608 (35.0)2667 (35.4)Vital signs Heart rate, beats/min86.0 (75.0–100.0)89.0 (77.0–104.0)85.0 (74.0–100.0)84.0 (74.0–98.0)84.0 (73.0–98.0) < 0.001 Systolic BP, mmHg123.0 (108.0–140.0)110.0 (98.0–139.0)121.0 (106.0–142.0)126.0 (112.0–142.0)128.0 (118.0–138.0) < 0.001 Diastolic BP, mmHg67.0 (57.0–79.0)65.0 (55.0–78.0)67.0 (56.0–79.0)67.0 (57.0–79.0)69.0 (60.0–80.0) < 0.001 Mean arterial pressure, mmHg83.0 (72.0–96.0)79.0 (68.0–94.0)83.0 (72.0–96.0)84.0 (74.0–96.0)86.0 (77.0–95.0) < 0.001Laboratory examinations Potassium, mEq/L4.1 (3.8–4.5)4.1 (3.7–4.6)4.1 (3.8–4.5)4.1 (3.8–4.5)4.1 (3.8–4.5) < 0.001 Sodium, mEq/L139.0 (136.0–141.0)138.0 (135.0–141.0)139.0 (136.0–141.0)139.0 (136.0–141.0)139.0 (136.0–141.0) < 0.001 Total calcium, mg/dL8.5 (8.0–9.0)8.4 (7.8–8.9)8.4 (7.9–8.9)8.5 (8.0–9.0)8.6 (8.1–9.0) < 0.001 Creatinine, mg/dL0.9 (0.7–1.3)1.0 (0.7–1.5)1.0 (0.7–1.4)0.9 (0.7–1.3)0.9 (0.7–1.2) < 0.001 BUN, mg/dL18.0 (13.0–28.0)19.0 (13.0–32.0)18.0 (13.0–29.0)18.0 (13.0–27.0)16.0 (12.0–24.0) < 0.001 AST, IU/L33.0 (21.0–67.0)36.0 (22.0–81.0)34.0 (21.0–70.0)31.0 (21.0–60.0)31.0 (20.0–61.0) < 0.001 ALT, IU/L25.0 (15.0–48.0)26.0 (15.0–53.0)25.0 (15.0–50.0)24.0 (15.0–44.5)25.0 (16.0–45.0) < 0.001 INR1.2 (1.1–1.4)1.2 (1.1–1.5)1.2 (1.1–1.4)1.2 (1.1–1.4)1.2 (1.1–1.4) < 0.001 Glucose, mg/dL124.0 (103.0–157.0)121.0 (100.0–155.0)125.0 (103.0–159.0)125.0 (104.0–159.0)124.0 (104.0–154.0) < 0.001 TG, mg/dL123.0 (85.0–190.0)119.0 (83.0–186.0)128.0 (86.0–198.0)124.0 (86.0–189.0)120.0 (84.0–185.0)0.014 TC, mg/dL148.0 (114.0–183.0)141.0 (105.0–176.0)147.0 (115.0–182.0)151.0 (118.0–185.0)153.0 (119.2–186.0) < 0.001 HDL, mg/dL42.0 (32.0–53.0)40.0 (30.0–52.0)42.0 (31.0–54.0)43.0 (33.0–54.0)42.0 (32.0–54.0) < 0.001 LDL, mg/dL78.0 (54.0–107.0)75.0 (49.0–105.0)76.0 (54.0–106.0)80.0 (56.0–109.0)80.0 (56.0–109.0)0.004 Lactate, mmol/L1.5 (1.1–2.3)1.7 (1.2–2.6)1.6 (1.1–2.3)1.5 (1.1–2.1)1.4 (1.1–2.1) < 0.001 Albumin, g/dL3.3 (2.8–3.8)3.1 (2.6–3.6)3.2 (2.8–3.7)3.4 (2.9–3.8)3.4 (3.0–3.9) < 0.001 CRP, mg/L72.3 (18.8–149.2)71.7 (24.5–149.7)76.3 (25.4–152.9)71.5 (15.9–154.6)59.8 (12.1–137.2)0.004 WBC, K/uL10.0 (7.3–14.0)10.2 (7.2–14.4)10.1 (7.2–14.2)10.0 (7.3–13.8)9.9 (7.3–13.5)0.016 RBC, K/uL3.8 (3.2–4.3)3.7 (3.1–4.2)3.7 (3.2–4.3)3.8 (3.3–4.3)3.9 (3.3–4.4) < 0.001 Platelet, K/uL200.0 (148.0–261.0)198.0 (142.0–263.0)199.0 (146.0–261.0)201.0 (151.0–262.0)201.0 (151.0–260.0)0.001 Hemoglobin, g/dL11.3 (9.6–12.9)11.0 (9.4–12.7)11.2 (9.6–12.8)11.4 (9.7–13.0)11.5 (9.8–13.1) < 0.001Comorbidities Coronary artery disease, n (%)14,470 (33.9)3180 (31.9)4366 (36.4)4656 (35.4)2268 (30.1) < 0.001 Atrial fibrillation, n (%)11,409 (26.8)2764 (27.7)3411 (28.5)3559 (27.0)1675 (22.2) < 0.001 Hypertension, n (%)27,216 (63.8)5755 (57.7)7922 (66.1)8913 (67.7)4626 (61.4) < 0.001 Heart failure, n (%)9713 (22.8)2786 (27.9)2757 (23.0)2812 (21.4)1358 (18.0) < 0.001 Diabetes mellitus, n (%)12,204 (28.6)2666 (26.7)3554 (29.7)3969 (30.2)2015 (26.7) < 0.001 Cerebrovascular disease, n (%)7266 (17.0)1328 (13.3)1883 (15.7)2546 (19.3)1509 (20.0) < 0.001 Chronic pulmonary disease, n (%)9931 (23.3)2374 (23.8)2887 (24.1)3083 (23.4)1587 (21.0) < 0.001 Renal disease, n (%)7702 (18.1)2008 (20.1)2263 (18.9)2310 (17.6)1121 (14.9) < 0.001 Tumors, n (%)6331 (14.8)1764 (17.7)1758 (14.7)1776 (13.5)1033 (13.7) < 0.001 Sepsis, n (%)7171 (16.8)2485 (24.9)2262 (18.9)1735 (13.2)689 (9.1) < 0.001Treatments Vasoactive agents, n (%)12,841 (30.1)3659 (36.7)4445 (37.1)3545 (26.9)1192 (15.8) < 0.001 Mechanical circulatory support, n (%)591 (1.4)219 (2.2)166 (1.4)124 (0.9)82 (1.1) < 0.001Outcomes In-hospital mortality, n (%)4481 (10.5)1465 (14.7)1335 (11.1)1178 (9.0)503 (6.7) < 0.001 Ventricular arrhythmias, n (%)3859 (9.0)1092 (11.0)1125 (9.4)1148 (8.7)494 (6.6) < 0.001ALT: alanine aminotransferase; AST: aspartate aminotransferase; BMI: body mass index; BP: blood pressure; BUN: blood urea nitrogen; CRP: C-reactive protein; HDL: high-density lipoprotein; ICU: intensive care unit; INR: international normalized ratio; LDL: low-density lipoprotein; RBC: red blood cell count; TC: total cholesterol; TG: triglyceride; WBC: white blood cell count Baseline characteristics of ICU patients according to quartile of SBP TTR ALT: alanine aminotransferase; AST: aspartate aminotransferase; BMI: body mass index; BP: blood pressure; BUN: blood urea nitrogen; CRP: C-reactive protein; HDL: high-density lipoprotein; ICU: intensive care unit; INR: international normalized ratio; LDL: low-density lipoprotein; RBC: red blood cell count; TC: total cholesterol; TG: triglyceride; WBC: white blood cell count Higher SBP TTR is associated with lower in-hospital mortality in ICU patients When analyzed as a continuous variable, each 10% increase in SBP TTR was associated with a 6–11% reduction in the odds of in-hospital all-cause mortality across all models (Model I: OR = 0.89, 95% CI 0.88–0.90; Model II: OR = 0.89, 95% CI 0.88–0.90; Model III: OR = 0.94, 95% CI 0.93–0.95; all P < 0.001) (Table 2). Subgroup analyses stratified by age (< 65 vs. ≥ 65 years), sex, race, hypertension, diabetes, and sepsis showed that the association between SBP TTR and mortality was generally consistent across subgroups. Notably, a significant interaction was observed for diabetes (P for interaction < 0.05), indicating a stronger association in patients with diabetes, whereas no significant interactions were found for the other subgroup variables (P for interaction > 0.05 for all; Fig. 2C). Table 2Association between SBP TTR and in-hospital all-cause mortality in ICU patientsModelModel 1Adjusted model IIAdjusted model IIIOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueContinuous variable0.89 (0.88, 0.90) < 0.0010.89 (0.88, 0.90) < 0.0010.94 (0.93, 0.96) < 0.001Quartile Q1111 Q20.73 (0.67, 0.79) < 0.0010.68 (0.63, 0.74) < 0.0010.80 (0.74, 0.88) < 0.001 Q30.57 (0.53, 0.62) < 0.0010.54 (0.50, 0.59) < 0.0010.74 (0.67, 0.81) < 0.001 Q40.41 (0.37, 0.46) < 0.0010.43 (0.38, 0.48) < 0.0010.66 (0.59, 0.75) < 0.001 P for Trend0.75 (0.73, 0.77) < 0.0010.75 (0.73, 0.77) < 0.0010.87 (0.84, 0.90) < 0.001Model I: unadjustedModel II: adjusted for age, sex, and BMIModel III: adjusted for age, sex, and BMI, race, ICU stay time, charlson comorbidity index, vital signs (heart rate, systolic BP, diastolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, hemoglobin, AST, ALT, albumin, TC, TG, HDL, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, mechanical circulatory support) Table 2Association between SBP TTR and in-hospital all-cause mortality in ICU patientsModelModel 1Adjusted model IIAdjusted model IIIOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueContinuous variable0.89 (0.88, 0.90) < 0.0010.89 (0.88, 0.90) < 0.0010.94 (0.93, 0.96) < 0.001Quartile Q1111 Q20.73 (0.67, 0.79) < 0.0010.68 (0.63, 0.74) < 0.0010.80 (0.74, 0.88) < 0.001 Q30.57 (0.53, 0.62) < 0.0010.54 (0.50, 0.59) < 0.0010.74 (0.67, 0.81) < 0.001 Q40.41 (0.37, 0.46) < 0.0010.43 (0.38, 0.48) < 0.0010.66 (0.59, 0.75) < 0.001 P for Trend0.75 (0.73, 0.77) < 0.0010.75 (0.73, 0.77) < 0.0010.87 (0.84, 0.90) < 0.001Model I: unadjustedModel II: adjusted for age, sex, and BMIModel III: adjusted for age, sex, and BMI, race, ICU stay time, charlson comorbidity index, vital signs (heart rate, systolic BP, diastolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, hemoglobin, AST, ALT, albumin, TC, TG, HDL, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, mechanical circulatory support) Association between SBP TTR and in-hospital all-cause mortality in ICU patients Model I: unadjusted Model II: adjusted for age, sex, and BMI Model III: adjusted for age, sex, and BMI, race, ICU stay time, charlson comorbidity index, vital signs (heart rate, systolic BP, diastolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, hemoglobin, AST, ALT, albumin, TC, TG, HDL, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, mechanical circulatory support) When SBP TTR was categorized into quartiles, a graded inverse relationship with in-hospital mortality was observed in all models. Compared with patients in the lowest quartile (Q1), those in Q2, Q3, and Q4 demonstrated progressively lower odds of mortality (Model I: ORs = 0.73, 0.57, 0.41; Model II: ORs = 0.68, 0.54, 0.43; Model III: ORs = 0.77, 0.70, 0.66; all P for trend < 0.001) (Table 2). RCS analysis was performed to test for nonlinearity (Fig. 2D). In all three models, the overall association between SBP TTR and mortality was significant (all P for overall < 0.001). In Models I and II, no evidence of nonlinearity was detected (P for nonlinearity = 0.191 and 0.063, respectively), whereas Model III showed a statistically significant nonlinearity (P for nonlinearity = 0.002), suggesting a more complex relationship after full adjustment. Nevertheless, the overall association remained inversely correlated, with higher SBP TTR associated with lower mortality. Sensitivity analyses further supported the robustness of these findings. When SBP TTR was calculated using the proportion method, or when covariates with more than 25% missing data were excluded, the inverse association between SBP TTR and in-hospital mortality remained consistent (Supplementary Tables 3, 4). Higher SBP TTR is associated with lower risk of VA in ICU patients When analyzed as a continuous variable, each 10% increase in SBP TTR was associated with a 3–7% reduction in the odds of VA across all models (Model I: OR = 0.93, 95% CI 0.92–0.95; Model II: OR = 0.93, 95% CI 0.92–0.94; Model III: OR = 0.97, 95% CI 0.95–0.98; all P < 0.001) (Table 3). Subgroup analyses stratified by age (< 65 vs. ≥ 65 years), sex, race, hypertension, diabetes, and sepsis revealed no significant interactions (P for interaction > 0.05 for all; Fig. 3A), suggesting the association between SBP TTR and VA was consistent across diverse clinical subgroups. Table 3Association between SBP TTR and VA in ICU patientsModelModel 1Adjusted model IIAdjusted model IIIOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueContinuous variable0.93 (0.92, 0.95) < 0.0010.93 (0.92, 0.94) < 0.0010.97 (0.95, 0.98) < 0.001Quartile Q1111 Q20.84 (0.77, 0.92) < 0.0010.77 (0.71, 0.84) < 0.0010.85 (0.78, 0.94)0.002 Q30.78 (0.71, 0.85) < 0.0010.72 (0.66, 0.78) < 0.0010.87 (0.78, 0.96)0.004 Q40.57 (0.51, 0.64) < 0.0010.58 (0.52, 0.65) < 0.0010.79 (0.70, 0.89) < 0.001 P for Trend0.85 (0.82, 0.88) < 0.0010.85 (0.82, 0.87) < 0.0010.93 (0.90, 0.97) < 0.001Model I: unadjustedModel II: adjusted for age, sex, and BMIModel III: adjusted for age, sex, and BMI, race, ICU stay time, charlson comorbidity index, vital signs (heart rate, systolic BP, diastolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, hemoglobin, AST, ALT, albumin, TC, TG, HDL, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, mechanical circulatory support)Fig. 3Association between SBP TTR and VA. A Subgroup analyses of the association between SBP TTR (per 10% increase) and risk of VA. ORs with 95% CIs were estimated using multivariable logistic regression. B RCS curves showing the inverse association between SBP TTR and VA. P values for overall and nonlinearity are shown Table 3Association between SBP TTR and VA in ICU patientsModelModel 1Adjusted model IIAdjusted model IIIOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueContinuous variable0.93 (0.92, 0.95) < 0.0010.93 (0.92, 0.94) < 0.0010.97 (0.95, 0.98) < 0.001Quartile Q1111 Q20.84 (0.77, 0.92) < 0.0010.77 (0.71, 0.84) < 0.0010.85 (0.78, 0.94)0.002 Q30.78 (0.71, 0.85) < 0.0010.72 (0.66, 0.78) < 0.0010.87 (0.78, 0.96)0.004 Q40.57 (0.51, 0.64) < 0.0010.58 (0.52, 0.65) < 0.0010.79 (0.70, 0.89) < 0.001 P for Trend0.85 (0.82, 0.88) < 0.0010.85 (0.82, 0.87) < 0.0010.93 (0.90, 0.97) < 0.001Model I: unadjustedModel II: adjusted for age, sex, and BMIModel III: adjusted for age, sex, and BMI, race, ICU stay time, charlson comorbidity index, vital signs (heart rate, systolic BP, diastolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, hemoglobin, AST, ALT, albumin, TC, TG, HDL, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, mechanical circulatory support) Association between SBP TTR and VA in ICU patients Model I: unadjusted Model II: adjusted for age, sex, and BMI Model III: adjusted for age, sex, and BMI, race, ICU stay time, charlson comorbidity index, vital signs (heart rate, systolic BP, diastolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, hemoglobin, AST, ALT, albumin, TC, TG, HDL, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, mechanical circulatory support) Association between SBP TTR and VA. A Subgroup analyses of the association between SBP TTR (per 10% increase) and risk of VA. ORs with 95% CIs were estimated using multivariable logistic regression. B RCS curves showing the inverse association between SBP TTR and VA. P values for overall and nonlinearity are shown When SBP TTR was analyzed as quartiles, a graded inverse association was observed in all models. Compared with patients in the lowest quartile (Q1), those in Q2, Q3, and Q4 demonstrated progressively lower odds of VA (Model I: ORs = 0.84, 0.78, 0.57; Model II: ORs = 0.77, 0.72, 0.58; Model III: ORs = 0.84, 0.85, 0.78; all P for trend < 0.001) (Table 3). RCS analysis further confirmed a significant overall association between SBP TTR and VA in all three models (all P for overall < 0.001; Fig. 3B). Evidence of nonlinearity was detected in the unadjusted model (Model I, P for nonlinearity = 0.002). However, after covariate adjustment, no significant nonlinearity was observed (P for nonlinearity = 0.056 and 0.074 in Models II and III, respectively), supporting an approximately linear inverse relationship in the fully adjusted models. Sensitivity analyses produced results consistent with the main analyses, supporting the robustness and stability of the association (Supplementary Table 3, 4 and 5). VA mediated the association between SBP TTR and in-hospital all-cause mortality The KHB mediation analysis indicated that VA mediated the relationship between SBP TTR and in-hospital all-cause mortality. The indirect effect accounted for only 1.96% of the total effect, suggesting that VA contributed minimally to the overall association (Table 4). Table 4VR as a mediator in the Relationship between SBP TTR and in-hospital all-cause mortalityEffectOR (95% CI)P valueDirect effect SBP TTR ≥ in-hospital all-cause mortality0.947 (0.933, 0.960) < 0.001Indirect effect SBP TTR ≥ in-hospital all-cause mortality (via VR)0.999 (0.998, 999)0.003 Total Effect0.946 (0.933, 0.959) < 0.001Proportion of Mediated = 1.95%. Adjusted for age, sex, and BMI, race, ICU stay time, charlson comorbidity index, vital signs (heart rate, systolic BP, diastolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, hemoglobin, AST, ALT, albumin, TC, TG, HDL, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, mechanical circulatory support) Table 4VR as a mediator in the Relationship between SBP TTR and in-hospital all-cause mortalityEffectOR (95% CI)P valueDirect effect SBP TTR ≥ in-hospital all-cause mortality0.947 (0.933, 0.960) < 0.001Indirect effect SBP TTR ≥ in-hospital all-cause mortality (via VR)0.999 (0.998, 999)0.003 Total Effect0.946 (0.933, 0.959) < 0.001Proportion of Mediated = 1.95%. Adjusted for age, sex, and BMI, race, ICU stay time, charlson comorbidity index, vital signs (heart rate, systolic BP, diastolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, hemoglobin, AST, ALT, albumin, TC, TG, HDL, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, mechanical circulatory support) VR as a mediator in the Relationship between SBP TTR and in-hospital all-cause mortality Proportion of Mediated = 1.95%. Adjusted for age, sex, and BMI, race, ICU stay time, charlson comorbidity index, vital signs (heart rate, systolic BP, diastolic BP, mean arterial pressure), laboratory examinations (potassium, sodium, total calcium, creatinine, BUN, WBC, RBC, platelet, hemoglobin, AST, ALT, albumin, TC, TG, HDL, LDL, glucose, INR, lactate, CRP), comorbidities (cerebrovascular disease, hypertension, diabetes, atrial fibrillation, coronary artery disease, heart failure, chronic pulmonary disease, renal disease, sepsis, tumors), and treatment (vasoactive agents, mechanical circulatory support) Predictive value of SBP TTR Independent of SBP CV After further adjustment for SBP CV, the inverse associations between SBP TTR and both in-hospital mortality (OR = 0.95, 95% CI 0.93–0.96; P < 0.001) and VA (OR = 0.97, 95% CI 0.95–0.98; P < 0.001) remained significant, indicating that SBP TTR provides prognostic information independent of blood pressure variability (Supplementary Table 6). ROC analyses demonstrated that SBP TTR and SBP CV had similar discriminative abilities for predicting in-hospital mortality (AUC = 0.59 for both) and VA (AUC = 0.55 for both), and both outperformed models based on SBP or MBP alone. Notably, combining SBP TTR and SBP CV further improved discrimination, yielding higher AUCs for in-hospital mortality (AUC = 0.62) and VA (AUC = 0.57), indicating that these two metrics provide complementary prognostic information (Supplement Fig. 1A, B). External validation in eICU-CRD database consistent with MIMIC findings The patient selection flowchart is shown in Supplementary Fig. 2A. A total of 55,155 ICU patients from the validation cohort were included. The median age was 65.0 years, 45.2% were female, and 78.3% were White (Supplementary Table 7). Overall, 10.6% experienced in-hospital mortality. Patients in the highest SBP TTR quartile (Q4) had the lowest incidence of mortality (Supplementary Fig. 2B), consistent with the primary cohort. In the validation cohort, higher SBP TTR was consistently associated with lower odds of in-hospital mortality across all models. When analyzed as a continuous variable, each 10% increase in SBP TTR was associated with a 6–10% reduction in the odds of mortality (Model I: OR = 0.90, 95% CI 0.89–0.91; Model II: OR = 0.90, 95% CI 0.89–0.91; Model III: OR = 0.94, 95% CI 0.93–0.95; all P < 0.001) (Supplementary Table 8). A graded inverse association was observed across SBP TTR quartiles, with patients in the highest quartile (Q4) showing the lowest mortality risk (all P for trend < 0.001), consistent with the findings from the primary cohort. RCS analysis in the validation cohort also suggested an inverse association between SBP TTR and in-hospital mortality, with no evidence of nonlinearity (P for nonlinearity > 0.05) (Supplementary Fig. 2C). Discussion In this study, we evaluated the association between SBP TTR and clinical outcomes for the first time in 42,648 ICU patients from the MIMIC-IV database. The results demonstrated that higher SBP TTR was associated with progressively lower rates of in-hospital all-cause mortality and VA, and these associations remained significant after multivariable adjustment. When SBP TTR was analyzed by quartiles, patients in the highest quartile exhibited the lowest risk of mortality. Notably, we found that SBP TTR provided prognostic information independent of the SBP CV, and combining these two metrics further improved risk discrimination, indicating their complementary roles in assessing hemodynamic stability. Furthermore, the inverse association between SBP TTR and the primary outcome (in-hospital mortality) was successfully validated in an external cohort of 55,155 ICU patients from the eICU-CRD database, confirming the robustness of this prognostic signal. Our findings are consistent with the previous studies. Prospective cohort studies in hypertensive populations have shown that patients who maintain blood pressure within the target range for a higher proportion of time have significantly lower risks of cardiovascular events and all-cause mortality [20, 21]. Analyses incorporating blood pressure variability have further indicated that low TTR combined with high BPV predicts adverse outcomes [20]. Additionally, in patients with heart failure, SBP TTR has been shown to independently predict major cardiovascular events [22]. Compared with these studies, our work is the first to apply the TTR concept to ICU patients, demonstrating that higher SBP TTR is associated with lower in-hospital mortality. Moreover, we found that SBP TTR provides complementary prognostic information to SBP CV; even after adjusting for SBP CV, TTR remained a significant predictor, suggesting that TTR adds predictive value beyond blood pressure variability alone. SBP TTR reflects the extent to which a patient’s blood pressure remains within an “optimal” control range over a given period. A higher TTR generally indicates more stable hemodynamics, implying fewer abrupt blood pressure swings and reduced exposure to harmful hypotension or hypertension. Theoretically, such stability preserves cerebral autoregulation and maintains myocardial oxygen balance. Notably, our mediation analysis showed that VA accounted for only a very small proportion (approximately 1.96%) of the association between SBP TTR and mortality. This finding suggests that the survival benefit of higher SBP TTR is likely not primarily driven by the prevention of arrhythmias. Instead, we hypothesize that alternative mechanisms play a dominant role; specifically, sustained hemodynamic stability may improve outcomes by minimizing cumulative ischemic or reperfusion-related organ injury, although further physiological studies are needed to confirm this pathway. Although our findings suggest that SBP TTR is closely associated with outcomes in ICU patients, they should be interpreted with caution given the retrospective and observational design of this study. We do not propose the TTR range identified here as a therapeutic target. The current guideline recommendations for common acute conditions indicate that systolic blood pressure should be lowered to < 140 mmHg after acute hemorrhagic stroke, and that a MAP ≥ 65 mmHg is typically used as the hemodynamic goal in septic shock [5, 23, 24]. However, ICU populations are highly heterogeneous, and no universally accepted optimal SBP target exists [5]. Recent studies in general populations and cohort settings have calculated SBP TTR using various target ranges and consistently found that higher TTR is associated with better outcomes. For example, Li et al. analyzed data from the SPRINT trial, using a 3-month SBP target of 110–140 mmHg, and reported that TTR within this range was most strongly associated with reduced cardiovascular event risk [11] Wang et al., based on the Chinese Kailuan cohort, compared TTR calculated for SBP targets of 110–130 mmHg versus 120–140 mmHg, finding that the 110–130 mmHg range yielded a stronger inverse association with left ventricular hypertrophy risk. Another 24-h dynamic monitoring study used a 110–140 mmHg target range and observed that higher SBP TTR was linked to a significantly lower risk of heart failure [15, 25]. Collectively, these non-ICU studies have used SBP target ranges of 110–130, 120–140, and 110–140 mmHg. In the present study, we therefore chose the relatively broad 110–140 mmHg range to better reflect the stability of blood pressure control in ICU patients. While our results suggest that patients with SBP maintained within this range tend to have better outcomes, this should not be interpreted as evidence that active titration to this SBP range is advisable in the ICU. Instead, TTR should be viewed primarily as a prognostic indicator reflecting the quality of blood pressure control or hemodynamic stability, rather than as a new therapeutic target. Future prospective studies or randomized controlled trials are needed to determine whether intentional modification of SBP TTR can causally improve clinical outcomes. The strengths of this study are threefold. First, it includes a large sample size and represents the first application of SBP TTR in an ICU population. Second, TTR was externally validated using an independent multicenter cohort from the eICU-CRD database, enhancing the robustness and generalizability of the findings. Third, the SBP measurements used to derive TTR reflect real-world ICU practice, relying on routine intermittent monitoring that is easily obtainable at the bedside, making the metric clinically practical and readily applicable in everyday ICU care. However, several limitations should be acknowledged. First, blood pressure measurements were obtained from both invasive and noninvasive methods, and for invasive measurements, the specific arterial site (e.g., radial vs. femoral) was not available, which may introduce heterogeneity; nevertheless, external validation using the multicenter eICU-CRD database supports the robustness of our findings. Second, although we required at least 19 h of BP recordings to reduce error from infrequent monitoring, intermittent measurements remain inherently limited, and future work incorporating continuous arterial waveform data may improve TTR precision. Third, although we required at least 19 h of BP recordings to reduce error from infrequent monitoring, intermittent measurements remain inherently limited, and future work incorporating continuous arterial waveform data may improve TTR precision. Finally, VA were identified using monitor-derived automated algorithms, which lack secondary manual adjudication by cardiologists. Consequently, potential classification bias may arise from signal artifacts (e.g., motion interference) or reduced algorithm sensitivity during complex rhythm transitions. As with all observational studies, residual confounding cannot be fully excluded. Conclusion Higher SBP TTR during the first 24 h of ICU admission was independently and inversely associated with in-hospital mortality and ventricular arrhythmias. This metric provides information complementary to blood pressure variability. The association with in-hospital mortality was consistently validated in the multicenter eICU-CRD database, suggesting SBP TTR may serve as a useful biomarker for risk stratification in ICU patients. Supplementary Information Supplementary Material 1: Table 1. Assessment of Multicollinearity Among Baseline Data .Supplementary Material 2 : Supplementary Figure 1. ROC analyses of SBP TTR and SBP CV for predicting in-hospital mortality and VA. A In-hospital mortality. B VA.Supplementary Material 3 : Supplement Figure 2. Patient selection flowchart and association of SBP TTR with in-hospital mortality in the eICU cohort. A Flowchart of patient selection. B Incidence of in-hospital mortality across quartiles of SBP TTR. C RCS curves demonstrating the inverse association between SBP TTR and in-hospital mortality. Supplementary Material 4: Table 2. Assessment of Multicollinearity Among Covariates.Supplementary Material 5: Table 3. Association Between SBP TTR (Proportion Method) and Outcome.Supplementary Material 6: Table 4. Sensitivity analysis excluding variables with >25% missing data.Supplementary Material 7: Table 5. Association Between SBP TTR and Secondary Outcomes (FPVCs and VT/VF).Supplementary Material 8: Table 6. Association Between SBP TTR and Outcomes in Model III Further Adjusted for SBP CV.Supplementary Material 9: Table 7. Baseline characteristics of ICU patients according to quartiles of SBP TTR base on the eICU-CRD database.Supplementary Material 10: Table 8. Association between SBP TTR and in-hospital all-cause mortality in ICU patients base on the eICU-CRD database. Supplementary Material 1: Table 1. Assessment of Multicollinearity Among Baseline Data . Supplementary Material 2 : Supplementary Figure 1. ROC analyses of SBP TTR and SBP CV for predicting in-hospital mortality and VA. A In-hospital mortality. B VA. Supplementary Material 3 : Supplement Figure 2. Patient selection flowchart and association of SBP TTR with in-hospital mortality in the eICU cohort. A Flowchart of patient selection. B Incidence of in-hospital mortality across quartiles of SBP TTR. C RCS curves demonstrating the inverse association between SBP TTR and in-hospital mortality. Supplementary Material 4: Table 2. Assessment of Multicollinearity Among Covariates. Supplementary Material 5: Table 3. Association Between SBP TTR (Proportion Method) and Outcome. Supplementary Material 6: Table 4. Sensitivity analysis excluding variables with >25% missing data. Supplementary Material 7: Table 5. Association Between SBP TTR and Secondary Outcomes (FPVCs and VT/VF). Supplementary Material 8: Table 6. Association Between SBP TTR and Outcomes in Model III Further Adjusted for SBP CV. Supplementary Material 9: Table 7. Baseline characteristics of ICU patients according to quartiles of SBP TTR base on the eICU-CRD database. Supplementary Material 10: Table 8. Association between SBP TTR and in-hospital all-cause mortality in ICU patients base on the eICU-CRD database.