Abstract
Background
Plasma volume status (PVS) is an indicator of fluid status, and fluid balance is linked to the prognosis of critically ill patients. This analysis aimed to examine the association of estimated PVS (ePVS) with mortality risk among traumatic brain injury (TBI) patients.
Methods
This retrospective cohort analysis extracted TBI patient data from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database spanning 2008–2022. ePVS was calculated using the Duarte formula (Duarte-ePVS) and the Kaplan–Hakim formula (KH-ePVS), respectively. The associations of Duarte-ePVS and KH-ePVS with 30-day mortality risk among TBI patients were examined via Cox regression analyses. The non-linear associations of ePVS with 30-day mortality were analyzed via the restricted cubic splines (RCS) curve.
Results
The analysis cohort comprised 1,440 TBI patients, survival outcomes showed 247 (17.15%) deaths and 1,193 (82.85%) survivors at 30-day follow-up. A 1-SD (standard deviation) increase in KH-ePVS levels [HR (95%CI) = 1.17 (1.02–1.34)] and KH-ePVS levels ≥11.458 [HR = 1.43 (1.06–1.95)] were linked to an elevated risk of 30-day mortality. No significant relationships of high Duarte-ePVS levels and Duarte-ePVS levels ≥7.134 with 30-day mortality risk were found (P > 0.05). No significant nonlinear association between KH-ePVS levels and 30-day mortality risk was found (P > 0.05).
Conclusion
Elevated KH-ePVS levels were associated with an increased risk of short-term mortality among TBI patients, indicating a potential role for KH-ePVS levels in risk assessment.
Keywords: 30-day mortality, estimated plasma volume status, KH-ePVS, MIMIC-IV, traumatic brain injury
Introduction
Traumatic brain injury (TBI) refers to the structural or physiological damage to brain function caused by external forces, typically resulting from falls, motor vehicle collisions, or assaults (1). TBI is a major cause of death and disability, imposing a heavy disease burden worldwide (2, 3). TBI occurs in 50–60 million people each year, and each TBI death is associated with approximately 24 years of life lost (2, 4). TBI patients transferred to the intensive care unit (ICU) have a high incidence of complications and short-term mortality (5–7). The main risk factors for poor prognosis in TBI patients involve age, severity of initial injury, increased intracranial pressure, and secondary injuries (cerebral edema, impaired perfusion, and hypoxia) (1, 2). Although advances in disease care, identifying factors that influence prognosis may aid in disease risk stratification and individualized treatment.
Fluid balance is critical in the management of ICU patients. Fluid overload can lead to poorer survival outcomes in critically ill patients, while implementing interventions to restrict positive cumulative fluid balance can improve patient prognosis (8, 9). Plasma volume (PV) constitutes the liquid component of circulating blood and is critical for maintaining vascular tension, tissue perfusion, and inflammatory homeostasis (10). PV can reflect fluid balance status, and an increase in PV is usually correlated with fluid overload (11). Several studies have reported that high PV load may exacerbate brain damage among TBI patients through excessive activation of the renin-angiotensin-aldosterone system (12–14). PV is usually measured using tracer dilution techniques, but these methods are expensive, time-consuming, and complex (10). Estimated plasma volume status (ePVS) based on hemoglobin and hematocrit (Duarte-ePVS) (15), and Kaplan–Hakim ePVS (KH-ePVS) based on hematocrit, body weight, and gender were proposed (16). These ePVS values were found to have similar effects to PV measured by tracer dilution techniques (15, 16). A recent study reported that high Duarte-ePVS and KH-ePVS levels were related to an increased risk of short-term mortality among intracerebral hemorrhage patients (17). However, the influence of ePVS levels on the prognosis of TBI patients has not been reported. Thus, this analysis aimed to explore the association between ePVS levels and mortality risk among TBI patients based on large sample data from a public database to assist in the clinical management and risk stratification of TBI patients.
Methods
Study design and patients
Data utilized in this retrospective cohort analysis were sourced from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database covering 2008–2022. MIMIC-IV records real inpatient data from patients admitted to a tertiary academic medical center (Boston, MA, USA), with data collection beginning in 2008 and currently updated to 2022 (https://mimic.mit.edu/docs/iv/about/). MIMIC-IV database collects demographic information, laboratory measurements, medications, vital signs, and mortality information for ICU patients. Ethical approval for the MIMIC database was granted by the Institutional Review Boards of the Beth Israel Deaconess Medical Center and the Massachusetts Institute of Technology. No informed consent was required as all patient records were fully anonymized. The analysis cohort comprised TBI cases at first ICU admission. The following exclusion criteria were applied to patient selection: (1) age < 18 years old; (2) the length of ICU stay < 24 h; (3) those diagnosed with polycythemia; (4) those with missing hemoglobin data; (5) those with missing information on weight; and (6) those with missing data on survival status. TBI patients were identified in the MIMIC database using the International Classification of Diseases (ICD-9/10) codes (ICD-9: 850- to 854-; ICD-10: S06, S060- to S066-).
Outcomes
The outcome of this analysis was 30-day mortality, which refers to the death of a patient from any cause within 30 days after ICU admission. In this analysis, patient survival status was recorded up to 30 days after ICU admission, with in-hospital survival data obtained from hospital records and out-of-hospital survival data extracted from Social Security Administration records.
Calculation of ePVS
The exposure variable in this analysis was ePVS. The calculation of ePVS was carried out using the Strauss-derived Duarte formula (Duarte-ePVS) (15) and the Kaplan–Hakim formula (KH-ePVS) (16), respectively. Duarte-ePVS was calculated as:
Duarte-ePVS = 100 × [1 – hematocrit] / hemoglobin (g/dL) where hematocrit is a fraction. KH-ePVS was calculated as: KH-ePVS = [(aPV – iPV) / iPV] × 100% where aPV = [1 - hematocrit] × [a + (b × weight (kg))] and the parameters of a and b are: (males: a = 1,530, b = 41; females, a = 864, b = 47.9); iPV = c × weight (kg) and the parameter of c is: (males: c = 39; females: c = 40). In the analysis, the classification of Duarte-ePVS (< 7.134, ≥7.134) and KH-ePVS (< 11.458, ≥11.458) was determined according to the optimal cutoff value using the “surv_cutpoint” method in the “survminer” package.
Data collection
Initial ICU admission records of patients were collected, containing age, gender, race, fluid balance within 24 h, Sequential Organ Failure Assessment (SOFA), Glasgow Coma Scale (GCS) score, lesion type, heart failure, acute myocardial infarction (AMI), sepsis, acute kidney injury (AKI), hematological system diseases, weight, heart rate, oxyhemoglobin saturation (SpO2), systolic, diastolic, respiratory rate, temperature, white blood cell (WBC), platelet, hemoglobin, red blood cell distribution width (RDW), hematocrit, creatinine, international normalized ratio (INR), blood urea nitrogen (BUN), glucose, calcium, anion gap, ventilation, vasopressor, neurosurgical intervention, mannitol, diuretic, length of ICU stay, and 30-day follow-up time. The 24-h fluid balance was calculated as 24-h fluid intake minus 24-h fluid output. AMI, heart failure, and hematological system diseases were determined via ICD-9/10 codes. Sepsis diagnosis followed Sepsis 3.0 criteria, requiring both suspected infection and a SOFA score ≥2 (18). AKI diagnosis followed KDIGO criteria (19). Polycythemia diagnostic criteria required hemoglobin >16.0 g/dL (females) or >16.5 g/dL (males), alternatively hematocrit exceeding 48% (females) or 49% (males) (20). Neurosurgical intervention included craniotomy and minimally invasive surgery.
Statistical analysis
Continuous variables' normality was assessed using skewness and kurtosis measures, with variance homogeneity checked by Levene's test. Continuous variables were shown as mean ± SD (standard deviation) or median and quartiles [M (Q1, Q3)], with group comparisons performed using t-test, t'-test, or Wilcoxon rank-sum test where applicable. Categorical variables were presented as counts and percentages [n (%)], with group comparisons conducted utilizing the Chi-square test or the Fisher exact test. Covariates with ≥20% missing data were removed, whereas those with < 20% underwent missing value imputation using random forest interpolation. Comparative assessments of variable differences before vs. after data imputation were conducted (Supplementary Table 1, Supplementary Figure 1).
Confounders related to 30-day mortality were first screened via univariable Cox analysis, then bidirectional stepwise regression applying Akaike Information Criterion (AIC) minimization criteria was utilized to screen for variables with P < 0.05 in univariable analysis (Supplementary Table 2). The associations of Duarte-ePVS and KH-ePVS with 30-day mortality among TBI patients were evaluated via univariable and multivariable Cox regression analyses, with associations presented as hazard ratio and 95% confidence interval [HR (95%CI)]. Age, race, SOFA, GCS, hematological system diseases, heart rate, temperature, WBC, RDW, BUN, glucose, ventilation, vasopressor, and mannitol were adjusted in the multivariable analyses. The non-linear associations of Duarte-ePVS and KH-ePVS with 30-day mortality were investigated via the restricted cubic splines (RCS) curve. The Kaplan-Meier (KM) curves were employed to evaluate the influence of Duarte-ePVS and KH-ePVS on 30-day survival. Furthermore, additional analyses of the ePVS and 30-day mortality association were conducted in age, gender, heart failure, sepsis, AKI, and hematological system diseases subgroups. R 4.4.3 software (Vienna, Austria) was employed for all analyses, with P < 0.05 considered significant.
Results
Patient characteristics
A total of 2,821 patients with TBI were identified from the MIMIC-IV database. After applying exclusion criteria, 1,381 patients were removed, leaving 1,440 eligible patients for final analysis (Figure 1). Comparison of characteristics between the included and excluded cohorts revealed significant differences across multiple variables (Supplementary Table 3). Among 1,440 enrolled TBI patients, survival outcomes showed 247 (17.15%) deaths and 1,193 (82.85%) survivors at 30-day follow-up. Table 1 lists the characteristics of TBI patients. Population characteristics showed 762 (52.92%) patients aged ≥65 years, 901 (62.57%) males, and 872 (60.56%) patients of White race. The mean values for Duarte-ePVS and KH-ePVS were 6.01 (±1.79) and 1.22 (±11.54), respectively. The median length of ICU stay was 2.86 (1.74, 5.95) days. The 30-day mortality group presented significant differences from 30-day survival group in terms of age, race, SOFA, GCS score, heart failure, sepsis, AKI, hematological system diseases, weight, heart rate, diastolic, temperature, WBC, platelet, hemoglobin, RDW, hematocrit, creatinine, INR, BUN, glucose, anion gap, ventilation, vasopressor, mannitol, diuretic, length of ICU stay, 30-day follow-up time, Duarte-ePVS, and KH-ePVS (P < 0.05).
Figure 1.
Screening flowchart for TBI patients. TBI, traumatic brain injury; ICU, intensive care unit; MIMIC-IV, the medical information mart for intensive care-IV.
Table 1.
Characteristics of patients with traumatic brain injury (TBI).
| Variables | Total (N = 1,440) | 30-day follow-up | P | |
|---|---|---|---|---|
| Survivors (N = 1193) | Non-survivors (N = 247) | |||
| Age, years, n (%) | < 0.001 | |||
| < 65 | 678 (47.08) | 602 (50.46) | 76 (30.77) | |
| ≥65 | 762 (52.92) | 591 (49.54) | 171 (69.23) | |
| Gender, n (%) | 0.191 | |||
| Male | 901 (62.57) | 756 (63.37) | 145 (58.70) | |
| Female | 539 (37.43) | 437 (36.63) | 102 (41.30) | |
| Race, n (%) | < 0.001 | |||
| White | 872 (60.56) | 741 (62.11) | 131 (53.04) | |
| Black | 79 (5.49) | 71 (5.95) | 8 (3.24) | |
| Others | 179 (12.43) | 159 (13.33) | 20 (8.10) | |
| Unknown | 310 (21.53) | 222 (18.61) | 88 (35.63) | |
| Fluid balance within 24 h, n (%) | 0.703 | |||
| ≤ 0 | 299 (20.76) | 245 (20.54) | 54 (21.86) | |
| >0 | 1,141 (79.24) | 948 (79.46) | 193 (78.14) | |
| SOFA, score, Mean (±SD) | 3.56 (±2.58) | 3.23 (±2.35) | 5.14 (±3.04) | < 0.001 |
| GCS, n (%) | 0.003 | |||
| ≥13 | 1,106 (76.81) | 935 (78.37) | 171 (69.23) | |
| < 13 | 334 (23.19) | 258 (21.63) | 76 (30.77) | |
| Lesion type, n (%) | 0.163 | |||
| Traumatic subdural hemorrhage | 464 (32.22) | 388 (32.52) | 76 (30.77) | |
| Traumatic subarachnoid hemorrhage | 211 (14.65) | 181 (15.17) | 30 (12.15) | |
| Traumatic subarachnoid hemorrhage/traumatic subdural hemorrhage | 169 (11.74) | 139 (11.65) | 30 (12.15) | |
| Focal brain injury | 139 (9.65) | 121 (10.14) | 18 (7.29) | |
| Intracranial injury of other or unspecified nature | 457 (31.74) | 364 (30.51) | 93 (37.65) | |
| Heart failure, n (%) | < 0.001 | |||
| No | 1,280 (88.89) | 1,082 (90.70) | 198 (80.16) | |
| Yes | 160 (11.11) | 111 (9.30) | 49 (19.84) | |
| AMI, n (%) | 0.193 | |||
| No | 1,402 (97.36) | 1,165 (97.65) | 237 (95.95) | |
| Yes | 38 (2.64) | 28 (2.35) | 10 (4.05) | |
| Sepsis, n (%) | 0.004 | |||
| No | 1,179 (81.88) | 993 (83.24) | 186 (75.30) | |
| Yes | 261 (18.12) | 200 (16.76) | 61 (24.70) | |
| AKI, n (%) | < 0.001 | |||
| No | 807 (56.04) | 697 (58.42) | 110 (44.53) | |
| Yes | 633 (43.96) | 496 (41.58) | 137 (55.47) | |
| Hematological system diseases, n (%) | 0.008 | |||
| No | 846 (58.75) | 720 (60.35) | 126 (51.01) | |
| Yes | 594 (41.25) | 473 (39.65) | 121 (48.99) | |
| Weight, kg, Mean (±SD) | 76.91 (±19.69) | 77.54 (±19.22) | 73.85 (±21.60) | 0.007 |
| Heart rate, bpm, Mean (±SD) | 85.41 (±18.75) | 84.83 (±18.12) | 88.20 (±21.37) | 0.022 |
| Systolic, mmHg, Mean (±SD) | 131.34 (±22.98) | 131.52 (±22.33) | 130.43 (±25.94) | 0.539 |
| Diastolic, mmHg, Mean (±SD) | 72.42 (±17.03) | 72.92 (±16.72) | 70.04 (±18.32) | 0.016 |
| Respiratory rate, bpm, Mean (±SD) | 18.60 (±4.87) | 18.53 (±4.74) | 18.96 (±5.42) | 0.245 |
| SPO2, %, Mean (±SD) | 97.65 (±3.02) | 97.61 (±2.95) | 97.86 (±3.31) | 0.229 |
| Temperature, °C, Mean (±SD) | 36.79 (±0.84) | 36.86 (±0.72) | 36.48 (±1.24) | < 0.001 |
| WBC, K/uL, M (Q1, Q3) | 10.50 (7.80, 13.70) | 10.30 (7.60, 13.50) | 11.90 (8.95, 16.20) | < 0.001 |
| Platelet, K/uL, M (Q1, Q3) | 189.00 (145.00, 241.00) | 190.00 (149.00, 242.00) | 177.00 (130.50, 230.50) | 0.025 |
| Hemoglobin, g/dL, Mean (±SD) | 11.47 (±2.01) | 11.59 (±1.95) | 10.89 (±2.17) | < 0.001 |
| RDW, %, Mean (±SD) | 14.15 (±1.82) | 13.98 (±1.68) | 14.94 (±2.21) | < 0.001 |
| Hematocrit, %, Mean (±SD) | 34.48 (±5.74) | 34.75 (±5.56) | 33.16 (±6.39) | < 0.001 |
| Creatinine, mg/dL, M (Q1, Q3) | 0.80 (0.70, 1.10) | 0.80 (0.70, 1.00) | 1.00 (0.70, 1.30) | < 0.001 |
| INR, ratio, Mean (±SD) | 1.23 (±0.24) | 1.21 (±0.22) | 1.32 (±0.31) | < 0.001 |
| BUN, mg/dL, Mean (±SD) | 18.31 (±11.87) | 17.24 (±10.69) | 23.49 (±15.43) | < 0.001 |
| Glucose, mg/dL, Mean (±SD) | 136.62 (±51.29) | 132.63 (±47.78) | 155.89 (±62.29) | < 0.001 |
| Calcium, mg/dL, Mean (±SD) | 8.42 (±0.76) | 8.43 (±0.73) | 8.34 (±0.88) | 0.125 |
| Anion gap, mEq/L, Mean (±SD) | 14.53 (±3.60) | 14.35 (±3.48) | 15.39 (±4.03) | < 0.001 |
| Ventilation, n (%) | < 0.001 | |||
| No | 304 (21.11) | 281 (23.55) | 23 (9.31) | |
| Yes | 1,136 (78.89) | 912 (76.45) | 224 (90.69) | |
| Vasopressor, n (%) | < 0.001 | |||
| No | 1,186 (82.36) | 1,030 (86.34) | 156 (63.16) | |
| Yes | 254 (17.64) | 163 (13.66) | 91 (36.84) | |
| Neurosurgical intervention, n (%) | 0.102 | |||
| No | 1,338 (92.92) | 1,102 (92.37) | 236 (95.55) | |
| Yes | 102 (7.08) | 91 (7.63) | 11 (4.45) | |
| Mannitol, n (%) | < 0.001 | |||
| No | 1,405 (97.57) | 1,176 (98.58) | 229 (92.71) | |
| Yes | 35 (2.43) | 17 (1.42) | 18 (7.29) | |
| Diuretic, n (%) | < 0.001 | |||
| No | 1,172 (81.39) | 1,000 (83.82) | 172 (69.64) | |
| Yes | 268 (18.61) | 193 (16.18) | 75 (30.36) | |
| Length of ICU stay, days, M (Q1, Q3) | 2.86 (1.74, 5.95) | 2.78 (1.70, 5.70) | 3.78 (2.05, 7.52) | < 0.001 |
| 30-days follow time, days, Mean (±SD) | 26.44 (±8.36) | 30.00 (±0.00) | 9.24 (±7.11) | < 0.001 |
| Duarte-ePVS, Mean (±SD) | 6.01 (±1.79) | 5.90 (±1.70) | 6.54 (±2.12) | < 0.001 |
| Duarte-ePVS, n (%) | < 0.001 | |||
| < 7.134 | 1,150 (79.86) | 982 (82.31) | 168 (68.02) | |
| ≥7.134 | 290 (20.14) | 211 (17.69) | 79 (31.98) | |
| KH-ePVS, Mean (±SD) | 1.22 (±11.54) | 0.51 (±11.06) | 4.62 (±13.14) | < 0.001 |
| KH-ePVS, n (%) | < 0.001 | |||
| < 11.458 | 1,192 (82.78) | 1,017 (85.25) | 175 (70.85) | |
| ≥11.458 | 248 (17.22) | 176 (14.75) | 72 (29.15) | |
SOFA, sequential organ failure assessment; GCS, Glasgow coma scale score; AMI, acute myocardial infarction; AKI, acute kidney injury; SPO2, oxyhemoglobin saturation; WBC, white blood cell; RDW, red blood cell distribution width; INR, international normalized ratio; BUN, blood urea nitrogen; ICU, intensive care unit; ePVS, estimated plasma volume status; Duarte-ePVS, ePVS based on hemoglobin and hematocrit (Duarte et al. formula); KH-ePVS, ePVS based on hematocrit, body weight, and gender (Kaplan–Hakim et al. formula)
Relationship between ePVS and 30-day mortality among TBI patients
Figure 2 displays the confounder-adjusted non-linear relationship between Duarte-ePVS, KH-ePVS, and 30-day mortality risk. No significant non-linear relationship was observed between Duarte-ePVS levels and 30-day mortality risk (Pnonlinear = 0.730, Poverall = 0.915; Figure 2A). The non-linear association between KH-ePVS and 30-day mortality risk was also not significant (Pnonlinear = 0.063, Poverall = 0.005). However, when KH-ePVS levels rise to a certain level, the 30-day mortality risk presents a rapid upward trend (Figure 2B).
Figure 2.
The non-linear relationship between ePVS and 30-day mortality risk among TBI patients. (A) Duarte-ePVS levels and 30-day mortality; (B) KH-ePVS levels and 30-day mortality. ePVS, estimated plasma volume status; TBI, traumatic brain injury; RCS, the restricted cubic splines.
The Cox regression analyses of the relationship between Duarte-ePVS, KH-ePVS, and 30-day mortality risk among TBI patients are shown in Table 2. A 1-SD increase in Duarte-ePVS levels [HR = 1.30 (1.17–1.44)] was associated with an elevated 30-day mortality risk in univariable analysis, and Duarte-ePVS levels ≥7.134 [HR = 1.95 (1.49–2.55)] also displayed a higher risk of 30-day mortality vs. Duarte-ePVS levels < 7.134. However, continuous Duarte-ePVS levels (P = 0.813) and Duarte-ePVS levels ≥7.134 (P = 0.449) were not associated with 30-day mortality risk in multivariable analyses. For KH-ePVS, high continuous KH-ePVS levels [unadjusted: HR = 1.37 (1.21–1.54); adjusted: HR = 1.19 (1.04–1.36)] and KH-ePVS levels ≥11.458 [unadjusted: HR = 2.10 (1.60–2.76); adjusted: HR = 1.47 (1.08–1.99)] were linked to an elevated risk of 30-day mortality both in univariable and multivariable analyses.
Table 2.
The Cox regression analyses of the relationship between ePVS and 30-day mortality risk among TBI patients.
| Variables | Univariable | Multivariable | ||
|---|---|---|---|---|
| HR (95% CI) | P | HR (95% CI) | P | |
| Duarte-ePVS | 1.30 (1.17–1.44) | < 0.001 | 1.02 (0.89–1.16) | 0.813 |
| Duarte-ePVS | ||||
| < 7.134 | Ref | Ref | ||
| ≥7.134 | 1.95 (1.49–2.55) | < 0.001 | 1.13 (0.82–1.55) | 0.449 |
| KH-ePVS | 1.37 (1.21–1.54) | < 0.001 | 1.19 (1.04–1.36) | 0.012 |
| KH-ePVS | ||||
| < 11.458 | Ref | Ref | ||
| ≥11.458 | 2.10 (1.60–2.76) | < 0.001 | 1.47 (1.08–1.99) | 0.015 |
TBI, traumatic brain injury; ePVS, estimated plasma volume status; Duarte-ePVS, ePVS based on hemoglobin and hematocrit (Duarte et al. formula); KH-ePVS, ePVS based on hematocrit, body weight, and gender (Kaplan–Hakim et al. formula); HR, hazard ratio; CI, confidence interval; Ref, reference; The multivariable Cox regression model was adjusted for age, race, SOFA, GCS, hematological system diseases, heart rate, temperature, WBC, RDW, BUN, glucose, ventilation, vasopressor, and mannitol.
A sensitivity analysis was performed by further adjusting for hematocrit, weight, and gender based on the original multivariable model (Supplementary Table 4). After additional adjustment for hematocrit, weight, and gender, continuous KH-ePVS levels [HR = 2.40 (1.59–3.61)] remained associated with the risk of 30-day mortality. However, the association between KH-ePVS levels ≥11.458 [HR = 1.46 (0.96–2.21), P = 0.073] and 30-day mortality risk was no longer statistically significant. Multicollinearity was assessed using variance inflation factor (VIF; Supplementary Table 5). In the Duarte-ePVS multivariable model, hematocrit (VIF = 10.97) and Duarte-ePVS (VIF = 11.63) exhibited severe multicollinearity. In the KH-ePVS multivariable model, moderate multicollinearity was observed for hematocrit (VIF = 8.81) and weight (VIF = 5.57), whereas KH-ePVS (VIF = 13.55) showed severe multicollinearity. These analyses indicated that derived variables (Duarte-ePVS or KH-ePVS) and their source variable (e.g., hematocrit, weight) should not be incorporated into the same model simultaneously.
Since many patients were excluded owing to an ICU stay of less than 24 h, a sensitivity analysis was conducted that included patients with an ICU stay exceeding 12 h (n = 1,779). High continuous KH-ePVS levels [HR = 1.19 (1.05–1.34)] and KH-ePVS levels ≥11.458 [HR = 1.44 (1.09–1.92)] were still associated with an increased risk of 30-day mortality (Supplementary Table 6).
Data from the MIMIC-III database (n = 438) were used to validate the association between ePVS and the risk of 30-day mortality (Supplementary Table 7). Duarte-ePVS levels ≥7.134 [unadjusted: HR = 1.56 (1.02–2.39)] and KH-ePVS levels ≥11.458 [unadjusted: HR = 1.73 (1.15–2.60)] were related to a higher risk of 30-day mortality in univariable analysis, but these associations were not significant in multivariable analysis (P > 0.05).
The influence of Duarte-ePVS and KH-ePVS on 30-day survival among TBI patients is shown in Figure 3. Both Duarte-ePVS levels ≥7.134 (vs. < 7.134) and KH-ePVS levels ≥11.458 (vs. < 11.458) presented poor survival probabilities among TBI patients (all Plog − rank < 0.0001). Moreover, patients in the KH-ePVS levels ≥11.458 group exhibited higher 24-h fluid balance values [1,325.25 (474.12, 2,835.11) vs. 1,043.84 (156.64, 2,229.88)] and a greater proportion of positive 24-h fluid balance (>0) [82.26 vs. 78.61%] compared with those in the KH-ePVS < 11.458 group (Supplementary Table 8).
Figure 3.
The Kaplan-Meier (KM) curves between ePVS and 30-day mortality among TBI patients. (A) Duarte-ePVS levels and 30-day mortality; (B) KH-ePVS levels and 30-day mortality. ePVS, estimated plasma volume status; TBI, traumatic brain injury.
The predictive performance of Duarte-ePVS and KH-ePVS for 30-day mortality risk was assessed. The Duarte-ePVS-combined SOFA [C-index: 0.694 (95%CI: 0.660–0.728)] and KH-ePVS-combined SOFA [C-index: 0.703 (95%CI: 0.670–0.736)] exhibited higher C-index values for predicting 30-day mortality risk than SOFA alone [C-index: 0.685 (95%CI: 0.651–0.720); Supplementary Table 9]. The Duarte-ePVS-combined SOFA [AUC: 0.713 (95%CI: 0.677–0.749)] and KH-ePVS-combined SOFA [AUC: 0.721 (95%CI: 0.686–0.756)] also showed higher AUC (the area under the receiver operating characteristic curve) values for predicting 30-day mortality risk than SOFA alone [AUC: 0.699 (95%CI: 0.662–0.736)] (Supplementary Table 10).
Figure 4 lists the association of KH-ePVS levels with 30-day mortality risk in different subgroups. These exploratory analyses suggest that KH-ePVS levels ≥11.458 exhibited a higher risk of 30-day mortality across subgroups: age [ < 65 years: HR = 1.85 (1.01–3.38); ≥65 years: HR = 1.47 (1.02–2.12)], males [HR = 1.65 (1.11–2.46)], non-heart failure [HR = 1.56 (1.10–2.20)], non-sepsis [HR = 1.67 (1.18–2.36)], non-AKI [HR = 2.36 (1.53–3.65)], and hematological system diseases [with: HR = 1.61 (1.08–2.41); without: HR = 1.89 (1.17–3.05)]. Furthermore, there may be an interaction between AKI and KH-ePVS levels on 30-day mortality risk (Pinteraction = 0.005).
Figure 4.
The association between KH-ePVS levels and 30-day mortality risk in different subgroups. ePVS, estimated plasma volume status; AKI, acute kidney injury.
Discussion
The association of ePVS with 30-day mortality risk among TBI patients was examined. Elevated KH-ePVS levels were linked to a higher 30-day mortality risk among TBI patients, but Duarte-ePVS levels were not significantly related to 30-day mortality risk. Sensitivity analyses indicated that the association between KH-ePVS and the risk of 30-day mortality was robust.
TBI caused by extracranial injury may result in systemic organ dysfunction, with systemic complications covering respiratory failure, renal failure, adrenal cortical insufficiency, myocardial injury, and potential multiple organ dysfunction (2). The ICU management for TBI primarily involves monitoring intracranial pressure and cerebral perfusion pressure (21). For ICU patients, fluid management is essential to restore cardiac output, systemic blood pressure, and renal perfusion. Fluid overload can cause complications such as pulmonary edema, cerebral edema, and increased intracranial pressure (22, 23). Fluid therapy using different intravenous infusions to maintain adequate organ tissue perfusion and oxygenation is critical for critically ill TBI patients (9). A prospective study indicated that a mean positive daily fluid balance (input > loss) was linked to higher ICU mortality and poorer functional recovery (9). The current analysis explored the association of ePVS with 30-day mortality risk among TBI patients. PV is related to the regulation between the interstitial and intravascular spaces and can be used as a marker of volume overload (24). Our results revealed that high KH-ePVS levels were associated with an increased 30-day mortality risk among TBI patients, but no significant association between Duarte-ePVS levels and 30-day mortality risk was found. Serum hemoglobin, blood glucose, base excess, and mean arterial pressure have been reported to influence early mortality in patients admitted to a neuro-ICU following TBI (25). Previous studies also reported that high ePVS levels were related to an elevated all-cause mortality risk in the general population (26) and short-term mortality risk in intracerebral hemorrhage patients (17). KH-ePVS and Duarte-ePVS represent two main PVS estimation formulas. Although good correlations between different formulas have been reported, there are differences in the correlations between different formulas and diseases (27–29). He et al. indicated that KH-ePVS is superior to Duarte-ePVS in predicting contrast-induced nephropathy among heart failure patients (27). Age plays a significant role in the prognostic management of patients with TBI (30). Due to age-related declines in physiological reserve, impaired fluid regulation, pre-existing chronic conditions, and long-term medication use, elderly TBI patients are more likely to present with abnormal laboratory findings upon admission (30); these factors may affect post-traumatic circulatory volume homeostasis. KH-ePVS levels may serve as a promising indicator for monitoring admission-related physiological abnormalities in elderly patients after TBI.
The mechanism by which elevated ePVS levels increase the short-term mortality risk among TBI patients remains unclear. Elevated ePVS indicates volume overload, which is caused by plasma volume expansion and can worsen cardiac function (24). High levels of ePVS may cause excessive activation of the renin-angiotensin-aldosterone system, which may result in cardiac remodeling and a subsequent decline in cardiac function (31, 32). Overactivation of neurohormones leads to water and sodium retention, which in turn causes hemodynamic congestion and extravascular edema (33, 34). Systemic congestion may be linked to progressive edema of brain tissue, which may lead to increased interstitial pressure and reduced collateral flow in the penumbra, ultimately resulting in collateral failure and hemorrhage in patients (12, 35). Fluid overload and increased central venous pressure may generate capillary hydrostatic pressure, leading to fluid accumulation in the brain stroma, which in turn increases traumatic cerebral edema and promotes disruption of the blood-brain barrier (9, 36). Volume overload is not only related to systemic congestion but also to atrial interference with inflamatosomes, which may lead to increased cerebral infarction volume and hemorrhage conversion (37). Animal studies have shown that volume overload is linked to increased oxidative stress, stiffness, and apoptosis in cardiomyocytes (38). Fluid overload increases arterial stiffness by elevating arterial dilation (39). Furthermore, complications caused by volume overload are also a major reason for increased mortality risk among TBI patients.
This analysis examined the correlation between ePVS calculated using different methods and the short-term mortality risk of TBI patients in the ICU. The sensitivity analysis further validated the reliability of the main findings of this study. However, validation using data from the MIMIC-III database revealed that the association between ePVS and the risk of 30-day mortality was not statistically significant in multivariate analysis. One possible explanation for the non-significant finding from multivariable analysis is the relatively small sample size of the validation cohort. Besides, inclusion of a wide array of confounders in the multivariable model may weaken statistical power, hindering the detection of associations linking exposure to outcome. Nevertheless, several limitations should be acknowledged. First, the analysis data are derived from retrospective data from a single center, which inevitably introduces selection bias that affects the results. Subsequent prospective data from multiple centers are needed to further validate the research results. Second, some disease characteristics, such as the location and severity of brain damage, cannot be obtained due to database limitations, which prevents analysis based on the location of brain damage and the severity of TBI. Third, the calculation of ePVS is based on the first measurement data within 24 h after admission to the ICU, which cannot reflect the changes of ePVS over time. Fourth, significant differences in multiple characteristics existed between included and excluded patients, which may introduce selection bias in the study population.
Conclusions
High KH-ePVS levels were linked to an increased risk of 30-day mortality among TBI patients, but not to Duarte-ePVS levels. Furthermore, it is necessary to further explore the prognostic predictive value of KH-ePVS for TBI patients, providing tools for risk stratification and prognostic management.
Funding Statement
The author(s) declared that financial support was received for this work and/or its publication. This study was supported by the Beijing Natural Science Foundation-Haidian Original Innovation Joint Fund Project (Grant No. L252077).
Footnotes
Edited by: Changming Wang, Capital Medical University, China
Reviewed by: Mario Ganau, Oxford University Hospitals NHS Trust, United Kingdom
Mingfeng Cao, Johns Hopkins University, United States
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Ethics statement
Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. Written informed consent from the patients/participants or patients/participants' legal guardian/next of kin was not required to participate in this study in accordance with the national legislation and the institutional requirements.
Author contributions
QQ: Writing – original draft, Writing – review & editing. WK: Writing – review & editing. QB: Writing – review & editing. YL: Writing – review & editing. XG: Writing – review & editing.
Conflict of interest
The handling editor CW declared a shared parent affiliation with the author(s) at the time of review.
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fneur.2026.1893304/full#supplementary-material
References
- 1.Blennow K, Brody DL, Kochanek PM, Levin H, McKee A, Ribbers GM, et al. Traumatic brain injuries. Nat Rev Dis Primers. (2016) 2:16084. doi: 10.1038/nrdp.2016.84 [DOI] [PubMed] [Google Scholar]
- 2.Maas AIR, Fitzgerald M, Gao G, Gupta D, Hutchinson P, Manley GT, et al. Traumatic brain injury: progress and challenges in prevention, clinical care, and research. Lancet Neurol. (2022) 21:1004–60. doi: 10.1016/S1474-4422(22)00307-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Global regional and and national burden of traumatic brain injury and spinal cord injury 1990-2016: 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol. (2019) 18:56–87. doi: 10.1016/s1474-4422(18)30415-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Majdan M, Plancikova D, Maas A, Polinder S, Feigin V, Theadom A, et al. Years of life lost due to traumatic brain injury in Europe: a cross-sectional analysis of 16 countries. PLoS Med. (2017) 14:e1002331. doi: 10.1371/journal.pmed.1002331 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Zhou YT, Tong DM, Wang SD, Ye S, Xu BW, Yang CX. Acute spontaneous intracerebral hemorrhage and traumatic brain injury are the most common causes of critical illness in the ICU and have high early mortality. BMC Neurol. (2018) 18:127. doi: 10.1186/s12883-018-1127-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Peng C, Yang F, Li L, Peng L, Yu J, Wang P, et al. A machine learning approach for the prediction of severe acute kidney injury following traumatic brain injury. Neurocrit Care. (2023) 38:335–44. doi: 10.1007/s12028-022-01606-z [DOI] [PubMed] [Google Scholar]
- 7.Qi L, Geng X, Feng R, Wu S, Fu T, Li N, et al. Association of glycemic variability and prognosis in patients with traumatic brain injury: a retrospective study from the MIMIC-IV database. Diabetes Res Clin Pract. (2024) 217:111869. doi: 10.1016/j.diabres.2024.111869 [DOI] [PubMed] [Google Scholar]
- 8.Malbrain ML, Marik PE, Witters I, Cordemans C, Kirkpatrick AW, Roberts DJ, et al. Fluid overload, de-resuscitation, and outcomes in critically ill or injured patients: a systematic review with suggestions for clinical practice. Anaesthesiol Intensive Ther. (2014) 46:361–80. doi: 10.5603/AIT.2014.0060 [DOI] [PubMed] [Google Scholar]
- 9.Wiegers EJA, Lingsma HF, Huijben JA, Cooper DJ, Citerio G, Frisvold S, et al. Fluid balance and outcome in critically ill patients with traumatic brain injury (CENTER-TBI and OzENTER-TBI): a prospective, multicentre, comparative effectiveness study. Lancet Neurol. (2021) 20:627–38. doi: 10.1016/s1474-4422(21)00162-9 [DOI] [PubMed] [Google Scholar]
- 10.Kalra PR, Anagnostopoulos C, Bolger AP, Coats AJ, Anker SD. The regulation and measurement of plasma volume in heart failure. J Am Coll Cardiol. (2002) 39:1901–08. doi: 10.1016/S0735-1097(02)01903-4 [DOI] [PubMed] [Google Scholar]
- 11.Shotan A, Dacca S, Shochat M, Kazatsker M, Blondheim DS, Meisel S. Fluid overload contributing to heart failure. Nephrol Dial Transplant. (2005) 20:vii24–27. doi: 10.1093/ndt/gfh1103 [DOI] [PubMed] [Google Scholar]
- 12.Zhang D, Li Q, Liu J, Ma L, Ye J, Hu G, et al. Calculated plasma volume status is associated with poor outcomes in acute ischemic stroke treated with endovascular treatment. Front Neurol. (2023) 14:1229331. doi: 10.3389/fneur.2023.1229331 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Villapol S, Janatpour ZC, Affram KO, Symes AJ. The renin angiotensin system as a therapeutic target in traumatic brain injury. Neurotherapeutics. (2023) 20:1565–91. doi: 10.1007/s13311-023-01435-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Li T, Huang HY, Wang HD, Gao CC, Liang H, Deng CL, et al. Restoration of brain angiotensin-converting enzyme 2 alleviates neurological deficits after severe traumatic brain injury via mitigation of pyroptosis and apoptosis. J Neurotrauma. (2022) 39:423–34. doi: 10.1089/neu.2021.0382 [DOI] [PubMed] [Google Scholar]
- 15.Duarte K, Monnez JM, Albuisson E, Pitt B, Zannad F, Rossignol P. Prognostic value of estimated plasma volume in heart failure. JACC Heart Fail. (2015) 3:886–93. doi: 10.1016/j.jchf.2015.06.014 [DOI] [PubMed] [Google Scholar]
- 16.Kobayashi M, Girerd N, Duarte K, Chouihed T, Chikamori T, Pitt B, et al. Estimated plasma volume status in heart failure: clinical implications and future directions. Clin Res Cardiol. (2021) 110:1159–72. doi: 10.1007/s00392-020-01794-8 [DOI] [PubMed] [Google Scholar]
- 17.Shen Y, Shen Y, Wang Y, Hu R, Xu H, Feng Y, et al. Associations of estimated plasma volume status with 30-day mortality and 1-year mortality in patients with intracerebral hemorrhage: a study of the MIMIC-IV database and the hospital information system. Front Neurol. (2025) 16:1548064. doi: 10.3389/fneur.2025.1548064 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The third international consensus definitions for sepsis and septic shock (sepsis-3). Jama. (2016) 315:801–10. doi: 10.1001/jama.2016.0287 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ostermann M, Bellomo R, Burdmann EA, Doi K, Endre ZH, Goldstein SL, et al. Controversies in acute kidney injury: conclusions from a kidney disease: improving global outcomes (KDIGO) conference. Kidney Int. (2020) 98:294–309. doi: 10.1016/j.kint.2020.04.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Randrianarisoa RMF, Ramanandafy H, Mania A, Monjanel H, Trouillier S. Prevalence and diagnostic performance of iron deficiency in polycythemia. Hematology. (2023) 28:2204621. doi: 10.1080/16078454.2023.2204621 [DOI] [PubMed] [Google Scholar]
- 21.Cnossen MC, Huijben JA, van der Jagt M, Volovici V, van Essen T, Polinder S, et al. Variation in monitoring and treatment policies for intracranial hypertension in traumatic brain injury: a survey in 66 neurotrauma centers participating in the CENTER-TBI study. Crit Care. (2017) 21:233. doi: 10.1186/s13054-017-1816-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Clifton GL, Miller ER, Choi SC, Levin HS. Fluid thresholds and outcome from severe brain injury. Crit Care Med. (2002) 30:739–45. doi: 10.1097/00003246-200204000-00003 [DOI] [PubMed] [Google Scholar]
- 23.Clifton GL, Valadka A, Zygun D, Coffey CS, Drever P, Fourwinds S, et al. Very early hypothermia induction in patients with severe brain injury (the National Acute Brain Injury Study: Hypothermia II): a randomised trial. Lancet Neurol. (2011) 10:131–9. doi: 10.1016/S1474-4422(10)70300-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Miller WL. Fluid volume overload and congestion in heart failure: time to reconsider pathophysiology and how volume is assessed. Circ Heart Fail. (2016) 9:e002922. doi: 10.1161/CIRCHEARTFAILURE.115.002922 [DOI] [PubMed] [Google Scholar]
- 25.Prisco L, Iscra F, Ganau M, Berlot G. Early predictive factors on mortality in head injured patients: a retrospective analysis of 112 traumatic brain injured patients. J Neurosurg Sci. (2012) 56:131–6. Available online at: https://www.minervamedica.it/en/journals/neurosurgical-sciences/article.php?cod=R38Y2012N02A0131# [PubMed] [Google Scholar]
- 26.Marawan A, Qayyum R. Estimated plasma volume and mortality: analysis from NHANES 1999-2014. Clin Res Cardiol. (2020) 109:1148–54. doi: 10.1007/s00392-020-01606-z [DOI] [PubMed] [Google Scholar]
- 27.He C, Zhang S, He H, You Z, Lin X, Zhang L, et al. Predictive value of plasma volume status for contrast-induced nephropathy in patients with heart failure undergoing PCI. ESC Heart Fail. (2021) 8:4873–81. doi: 10.1002/ehf2.13681 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Fudim M, Lerman JB, Page C, Alhanti B, Califf RM, Ezekowitz JA, et al. Plasma volume status and its association with in-hospital and postdischarge outcomes in decompensated heart failure. J Card Fail. (2021) 27:297–308. doi: 10.1016/j.cardfail.2020.09.478 [DOI] [PubMed] [Google Scholar]
- 29.Kobayashi M, Huttin O, Donal E, Duarte K, Hubert A, Le Breton H, et al. Association of estimated plasma volume status with hemodynamic and echocardiographic parameters. Clin Res Cardiol. (2020) 109:1060–69. doi: 10.1007/s00392-020-01599-9 [DOI] [PubMed] [Google Scholar]
- 30.Depreitere B, Becker C, Ganau M, Gardner RC, Younsi A, Lagares A, et al. Unique considerations in the assessment and management of traumatic brain injury in older adults. Lancet Neurol. (2025) 24:152–65. doi: 10.1016/S1474-4422(24)00454-X [DOI] [PubMed] [Google Scholar]
- 31.Pugliese NR, Masi S, Taddei S. The renin-angiotensin-aldosterone system: a crossroad from arterial hypertension to heart failure. Heart Fail Rev. (2020) 25:31–42. doi: 10.1007/s10741-019-09855-5 [DOI] [PubMed] [Google Scholar]
- 32.Chen J, Shen J, Cai D, Wei T, Qian R, Zeng C, et al. Estimated plasma volume status (ePVS) is a predictor for acute myocardial infarction in-hospital mortality: analysis based on MIMIC-III database. BMC Cardiovasc Disord. (2021) 21:530. doi: 10.1186/s12872-021-02338-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Grodin JL, Philips S, Mullens W, Nijst P, Martens P, Fang JC, et al. Prognostic implications of plasma volume status estimates in heart failure with preserved ejection fraction: insights from TOPCAT. Eur J Heart Fail. (2019) 21:634–42. doi: 10.1002/ejhf.1407 [DOI] [PubMed] [Google Scholar]
- 34.Gelman S, Mushlin PS. Catecholamine-induced changes in the splanchnic circulation affecting systemic hemodynamics. Anesthesiology. (2004) 100:434–9. doi: 10.1097/00000542-200402000-00036 [DOI] [PubMed] [Google Scholar]
- 35.Al-Dasuqi K, Payabvash S, Torres-Flores GA, Strander SM, Nguyen CK, Peshwe KU, et al. Effects of collateral status on infarct distribution following endovascular therapy in large vessel occlusion stroke. Stroke. (2020) 51:e193–202. doi: 10.1161/STROKEAHA.120.029892 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.McManus ML, Strange K. Acute volume regulation of brain cells in response to hypertonic challenge. Anesthesiology. (1993) 78:1132–7. doi: 10.1097/00000542-199306000-00017 [DOI] [PubMed] [Google Scholar]
- 37.Shen X, Zhang X, Liu M, Dong N, Liao J, Zhou G, et al. NT-proBNP levels and collateral circulation status in patients with acute ischemic stroke. Dis Markers. (2023) 2023:5318012. doi: 10.1155/2023/5318012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Mohamed BA, Schnelle M, Khadjeh S, Lbik D, Herwig M, Linke WA, et al. Molecular and structural transition mechanisms in long-term volume overload. Eur J Heart Fail. (2016) 18:362–71. doi: 10.1002/ejhf.465 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Tycho Vuurmans JL, Boer WH, Bos WW, Blankestijn PJ, Koomans HA. Contribution of volume overload and angiotensin II to the increased pulse wave velocity of hemodialysis patients. J Am Soc Nephrol. (2002) 13:177–83. doi: 10.1681/ASN.V131177 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.