INTRODUCTION
Traumatic brain injury (TBI) is a leading cause of morbidity and mortality globally, affecting millions of individuals each year and contributing substantially to long-term disability and health care burden. Traumatic intracranial hemorrhage (tICH) has been recognized as a common and serious type of TBI, including intraparenchymal hemorrhage, epidural hematoma, subdural hematoma, and subarachnoid hemorrhage. The mortality rate of tICH reaches up to 20-60%, even more than 60% in severe trauma patients [29]. Although the primary insult in tICH is neurological, the ensuing cascade of systemic responses often results in multi-organ dysfunction, complicating both acute management and long-term recovery [3,24]. Among extracerebral complications, acute kidney injury (AKI) is particularly concerning condition due to its high incidence, association with prolonged intensive care unit (ICU) stays, increased healthcare costs, and its strong correlation with short- and long-term mortality [9].
In critically ill patients, the kidneys are particularly vulnerable to episodes of hypo- and hyper-perfusion, and the ability to maintain stable renal blood flow is heavily influenced by the consistency of systemic blood pressure [18]. In tICH patients, the risk and severity of AKI may be even greater due to the intensified systemic inflammatory response, increased intracranial pressure (ICP), and the frequent need for interventions such as hyperosmolar therapy, vasopressors, or nephrotoxic agents [19]. These factors further deepen the vulnerability of renal function in that patient population and increase the need for early detection and individualized renal protection strategies [5,9].
Blood pressure variability (BPV), defined as the magnitude of fluctuations in blood pressure over a given period, has increasingly been recognized as an important hemodynamic parameter beyond absolute blood pressure values. Several statistical metrics have been developed to quantify BPV, with the most commonly used being standard deviation (SD), coefficient of variation (CV), and average real variability (ARV) [27]. These metrics offer complementary insights into the dynamic nature of blood pressure and its physiological impact.
An increasing number of studies have demonstrated that elevated BPV is associated with poor cardiovascular [26] and cerebrovascular [5,8] outcomes, as well as higher mortality among critically ill patients [27]. BPV is thought to impair end-organ perfusion by disrupting autoregulatory mechanisms and promoting endothelial dysfunction, oxidative stress, and inflammation, all of which may contribute to renal microvascular injury and the development of AKI [8,14].
In TBI, tight control of systemic blood pressure is essential to maintain cerebral perfusion and prevent secondary insults [13,16,30]. This is particularly important in tICH, as both elevated and reduced blood pressure can adversely affect clinical outcomes by influencing hematoma dynamics and cerebral perfusion. However, recent studies suggest that excessive fluctuations in blood pressure, even in the setting of acceptable average values, may predispose patients to renal hypoperfusion and ischemic injury [18,25]. These findings raise the question of current blood pressure management strategies and renal function protection in the treatment of neurological intensive care. Despite the increasing interest in modifiable risk factors affecting BPV, its clinical relevance to the development of AKI after tICH is not yet fully elucidated.
This study aims to investigate the relationship between BPV and the development of AKI in patients with tICH. By clarifying the potential role of BPV in the pathophysiology of AKI, the findings may inform hemodynamic management strategies to reduce renal complications and improve outcomes in this vulnerable population.
MATERIALS AND METHODS
Data source
The study data were taken from the Medical Information Mart for Intensive Care (MIMIC)-IV (version 2.2) database. MIMIC-IV is a comprehensive and publicly available database designed for research on critical care. Developed by the Institutional Review Boards of the Massachusetts Institute of Technology (MIT, Cambridge, MA, USA), MIMIC-IV contains deidentified health data associated with patients admitted to ICUs at the Beth Israel Deaconess Medical Center in Boston, Massachusetts [11]. To access the database, we passed the National Institutes of Health Protecting Human Research Participants webbased training course, and we obtained approval to extract data from MIMIC-IV for research purposes (certification number : 62708536). This study does not require Institutional Review Board (IRB) approval because dataset of MIMIC-IV has been deidentified.
Data extraction study population
Patients with tICH were identified in MIMIC-IV (version 2.2) database using International Classification of Diseases (ICD) diagnosis codes. Specifically, among ICD-9 codes 850-854 and ICD-10 codes S06.1-S06.9, we included only those cases that explicitly indicated the presence of hemorrhage (Fig. 1). Demographics, vital signs, laboratory tests, scoring systems, comorbidities [2], relevant treatment information, and others were extracted from MIMIC-IV database using structured query language with PostgreSQL (version 14; PostgreSQL Global Development Group, Berkeley, CA, USA).
Exclusion criteria
Patients were excluded from the study based on the following criteria : age under 18 years; length of stay in the ICU less than 1 day or greater than 30 days; patients who were not on their first ICU admission; individuals with chronic kidney disease or complicated diabetes mellitus; those without a diagnosis of intracranial hemorrhage; and patients with fewer than 24 blood pressure measurements on the first day of ICU admission (Fig. 1).
BPV
In clinical studies assessing variability in physiological or laboratory measurements, several statistical indices are commonly used, including SD, CV, and ARV [7]. To assess BPV, only patients with at least 24 blood pressure measurements within the first 24 hours were included.
ARV calculates the average of the absolute differences between consecutive blood pressure readings. It is useful for capturing short-term, beat-to-beat or hourly changes.
SD measures how much individual blood pressure readings deviate from the average value. A higher SD means greater variability over time.
CV is the SD divided by the mean blood pressure, usually expressed as a percentage. It reflects variability relative to the average pressure, making it easier to compare across individuals.
AKI definitions
The primary outcome of our study was the development of AKI during the ICU stay. Kidney Disease : Improving Global Outcomes (KDIGO) criteria were used to determine the occurrence of AKI. According to this definition, stage 1 is diagnosed when there is an increase in serum creatinine (S-cr) to 1.5 times the baseline value or urine output (UO) <0.5 mL/kg/h for 6-12 hours [12]. Stage 2 is defined by an increase in S-cr of 2.0-2.9 times the baseline level or UO <0.5 mL/kg/h for ≥12 hours. Stage 3 is defined as an increase in S-cr of 3.0 times the baseline value or UO <0.3 mL/kg/h for ≥24 hours or anuria for ≥12 hours. In this study, AKI was defined as KDIGO stage 1 or higher. Baseline serum creatinine was defined as the lowest value within 7 days before ICU admission.
Statistical analysis
Three types of BPV metrics were evaluated in this study : ARV, SD, and CV. These measures were each calculated for systolic blood pressure (SBP) and diastolic blood pressure (DBP), resulting in a total of six BPV indices (SBP_ARV, DBP_ARV; SBP_SD, DBP_SD; and SBP_CV, DBP_CV).
To evaluate their independent association with AKI, we first constructed univariable logistic regression models for each BPV metric. Subsequently, we performed multivariable logistic regression using six BPV metrics. Analyses were stratified by confounder adjustment groups : group 1 included demographic and clinical factors (age, congestive heart failure [CHF], and diabetes); group 2 included all factors from group 1 plus pharmacologic treatments (vasoactive agent use, mannitol, angiotensin-converting enzyme [ACE] inhibitors, angiotensin receptor blockers [ARBs], hypertonic saline). Odds ratios (OR) and 95% confidence intervals (CIs) were reported for each BPV parameter. In addition, analyses were conducted in the hypertensive subgroup, defined as patients with a mean SBP ≥140 mmHg during the first 24 hours of ICU admission, calculated as the average of all hourly SBP measurements recorded in the electronic database [10]. To ensure that BPV preceded the development of AKI and to address possible reverse causality, a sensitivity analysis was conducted after excluding patients who developed AKI within the first 24 hours of ICU admission.
All statistical analyses were performed using R version 2023.03.0+386 (R Foundation for Statistical Computing, Vienna, Austria), and a two-sided p-value <0.05 was considered statistically significant.
RESULTS
Patients’ characteristics
Among the total cohort of 551 patients included in the study, 385 individuals (69.9%) developed AKI. There were no significant differences in age, sex, mean systolic or DBP, or Glasgow coma scale scores between groups. However, patients with AKI had notably longer ICU (3.66 vs. 1.82 days, p<0.001) and hospital stays (7.89 vs. 4.9 days, p<0.001), and higher Simplified acute physiology score II (35 vs. 31, p<0.001). Among laboratory parameters, hemoglobin was modestly lower in the AKI group (p=0.033), while other markers such as white blood cell, platelet count, coagulation parameters, and renal markers (blood urea nitrogen, creatinine) showed no significant differences. AKI patients were more likely to have received vasopressors (29.6% vs. 11.4%, p<0.001), reflecting greater hemodynamic instability. The use of ACE inhibitors, ARBs, mannitol, and hypertonic saline did not differ significantly between the groups. Comorbidity profiles revealed a higher prevalence of CHF (p=0.007) and diabetes mellitus (p=0.018) in the AKI group (Table 1). For the sensitivity analysis, we excluded patients who developed AKI within the first 24 hours after ICU admission (n=241). After this exclusion, the cohort consisted of 310 patients, including 144 who developed AKI (46.5%) and 166 who did not (53.5%).
BPV
Analysis of BPV revealed significant differences between patients who developed AKI and those who did not. Specifically, systolic BPV was notably higher in patients with AKI than in those without AKI across multiple metrics, including SBP_ARV (11.3 vs. 9.98, p=0.002), SBP_SD (14.7 vs. 13.8, p<0.001), and SBP_CV (12.0 vs. 10.8, p<0.001). Diastolic BP variability also showed higher values in the AKI group, with statistically significant differences in DBP_SD (9.67 vs. 8.96, p=0.015) and DBP_CV (15.4 vs. 14.9, p=0.038), although DBP_ARV did not reach statistical significance (p=0.166). These findings suggest that increased BPV, particularly in systolic parameters, may be associated with the development of AKI in critically ill patients, supporting the prognostic relevance of dynamic blood pressure measures beyond mean values alone (Table 2).
Relationship between BPV and AKI in tICH patients
Univariate logistic regression analyses demonstrated that several BPV parameters were significantly associated with the risk of AKI. Notably, higher systolic BPV measures, including SBP_ARV (OR, 1.063; 95% CI, 1.020-1.09; p=0.004), SBP_SD (OR, 1.074; 95% CI, 1.033-1.119; p<0.001), and SBP_CV (OR, 1.089; 95% CI, 1.039-1.144; p<0.001), were all associated with increased odds of AKI. Among diastolic measures, DBP_SD (OR, 1.062; 95% CI, 1.010-1.120; p=0.023) and DBP_CV (OR, 1.034; 95% CI, 1.002-1.068; p=0.045) were also significant, while DBP_ARV was not (Supplementary Table 1).
After adjusting for confounders in multivariable logistic models, systolic BPV remained an independent predictor of AKI in both groups. In group G1, each 1-unit increase in SBP_ARV and SBP_SD was associated with 6.1% (OR, 1.061; 95% CI, 1.017-1.110; p=0.008) and 7.4% (OR, 1.074; 95% CI, 1.032-1.120; p<0.001) greater odds of AKI, respectively. DBP_SD (OR, 1.058; 95% CI, 1.006-1.117; p=0.034) and SBP_CV (OR, 1.088; 95% CI, 1.004-1.095; p<0.001) also showed significant associations. In the fully adjusted group G2, SBP_ARV and SBP_SD were associated with 4.6% (OR, 1.046; 95% CI, 1.004-1.095; p=0.048) and 5.2% (OR, 1.052; 95% CI, 1.010-1.099; p=0.018) increased odds of AKI, respectively, while no other BPV metric reached statistical significance (Table 3).
In a sensitivity analysis excluding patients who developed AKI within the first 24 hours after ICU admission, the associations between systolic BP variability indices (ARV, SD, and CV) and AKI remained directionally consistent, although the statistical significance was slightly attenuated (Supplementary Table 2).
Relationship between BPV and AKI in hypertensive tICH patients
Among the 551 patients, 63 (11.4%) had a SBP of ≥140 mmHg on admission, of whom 46 (73.0%) developed AKI. In the subgroup of patients with mean SBP ≥140 mmHg, SBP_ARV remained a significant predictor of AKI in group 1 (OR, 1.197; 95% CI, 1.010-1.418; p=0.038). In group 2, these associations were even stronger : SBP_ARV (OR, 1.285; 95% CI, 1.064-1.553; p=0.009), SBP_SD (OR, 1.203; 95% CI, 1.022-1.417; p=0.027) and SBP_CV (OR, 1.273; 95% CI, 1.012-1.601; p=0.039) all predicted higher AKI risk. No diastolic BPV metric was significant in either group (Table 4).
In the fully adjusted model (group 2), all SBP variability indices demonstrated positive associations with the risk of AKI. Compared with the overall cohort, the hypertensive subgroup showed higher ORs across all SBP variability measures. CIs were generally wider in the hypertensive subgroup, reflecting smaller sample size and reduced precision, despite larger effect sizes (Fig. 2).
DISCUSSION
The present study demonstrates, for the first time in a neurocritical care cohort, that elevated systolic BPV is an independent predictor of AKI in tICH patients, with the strongest associations observed in those with hypertension. These findings extend prior work linking BPV to adverse cardiovascular and cerebrovascular outcomes by identifying the kidney as an additional end‐organ vulnerable to hemodynamic instability. Collectively, our results underscore the need to supplement traditional mean arterial pressure targets with strategies aimed at mitigating extreme systolic pressure swings in the acute management of tICH.
Traditional clinical practice has focused on maintaining average blood pressure within a defined range. However, growing evidence suggests that variability around the mean may be equally, if not more, important in influencing outcomes such as cardiovascular events, mortality, and organ failure. BPV is thought to impair autoregulation, promote endothelial dysfunction, and contribute to oxidative stress and inflammation, all of which can play a role in the pathogenesis of AKI. In this study, higher values of SBP_ARV, SBP_SD, and SBP_CV were observed in patients who developed AKI, and these differences were statistically significant. These results are in line with previous studies in broader ICU populations, which have linked high BPV to increased risk of renal and cardiovascular complications [1,17,22,23].
In our analysis, the strength and significance of associations between BPV metrics and AKI differed by adjustment scheme and by the metric used. Each BPV index reflects a distinct aspect of blood pressure dynamics : SD is widely used for its simplicity but ignores the temporal sequence of measurements and is sensitive to outliers; CV normalizes variability to mean BP, enabling comparisons across patients with different baseline pressures but can be distorted when the mean BP is low; ARV incorporates the order of measurements and captures short-term fluctuations with less influence from extreme values. The finding that SBP_ARV and SBP_SD remained significant after full adjustment suggests that both overall dispersion and short-term systolic BP swings may be particularly harmful to renal function in tICH patients. These differences underscore the need to select BPV indices appropriate to the clinical context, while also considering that covariate structure, sample size in subgroups, and underlying vascular autoregulatory capacity may modulate observed associations. Importantly, these associations remained directionally consistent even after excluding patients who developed AKI within the first 24 hours after ICU admission, supporting the temporal sequence between elevated BP variability and subsequent renal injury and reducing the likelihood of reverse causality.
The interplay between cerebral and renal physiology in tICH patients is complex. Following tICH, sympathetic overactivity, catecholamine surges, and systemic inflammation are common and can lead to hemodynamic instability and multiorgan dysfunction. The kidneys are particularly vulnerable to changes in perfusion due to their reliance on stable autoregulation mechanisms [5]. Disruption of these mechanisms by high BPV can lead to repeated insults of ischemia and reperfusion, promoting tubular injury and progression to AKI. Although early AKI cases accounted for a notable proportion of events, they likely reflect both preexisting renal vulnerability and the immediate hemodynamic stress during early resuscitation, when abrupt BP fluctuations may precipitate renal hypoperfusion. Dysregulation of the renin-angiotensin-aldosterone system (RAAS) in this setting may further impair renal perfusion autoregulation, amplifying the adverse effects of BP variability. Once AKI occurs, the resulting fluid imbalance and neurohormonal activation can in turn exacerbate BP instability, suggesting a bidirectional interaction in which BPV and renal dysfunction reinforce each other. Thus, early AKI may arise from the synergistic effects of baseline renal fragility, RAAS-mediated hemodynamic dysregulation, and this vicious cycle between BP fluctuations and renal injury. Additionally, the inflammatory milieu that follows TBI may sensitize the kidneys to hemodynamic stress, compounding the effects of BPV. Our findings emphasize that BP management in tICH should not be limited to achieving a target mean BP but should also include efforts to minimize abrupt fluctuations that may challenge renal perfusion [30].
Tight SBP regulation is a cornerstone of neurocritical care in TBI, particularly in those with intracranial hemorrhage. The rationale is clear : both hypotension and hypertension can exacerbate secondary brain injury. Hypotension is a well-established risk factor for poor neurological outcomes, as it can compromise cerebral perfusion pressure (CPP) and precipitate cerebral ischemia, especially in the setting of impaired autoregulation following TBI [6]. Conversely, hypertension is associated with hematoma expansion in intracerebral hemorrhage, which can worsen mass effect and neurological deterioration [15]. For this reason, not only the absolute values of blood pressure but also the degree of BPV have emerged as critical determinants of outcome. Recent studies suggest that excessive BPV, even when mean SBP is maintained within target ranges, may be independently associated with worse neurological outcomes in TBI patients [28]. Fluctuations in blood pressure can further disrupt cerebral autoregulation, potentially leading to periods of hypoperfusion or rebleeding. Nevertheless, studies on BPV in tICH are insufficient, leaving uncertainty about its role as a modifiable therapeutic target. Our findings show prognostic relevance of BPV in this population and suggest implications for acute-stage management.
In our subgroup analysis, we found that SBP_ARV showed a similar or slightly more pronounced association with AKI among patients with a mean systolic BP ≥140 mmHg, although the CIs were wide due to the small sample size. This observation may reflect underlying vascular dysfunction or impaired autoregulatory capacity in patients with pre-existing or stress-induced hypertension [21]. In such patients, elevated BPV could more easily lead to episodes of renal dysfunction [4], disrupting glomerular filtration and promoting ischemic injury [20]. This is particularly relevant in tICH patients, who often require tight hemodynamic control to preserve cerebral perfusion. Clinicians may use vasopressors or fluid resuscitation to target CPP, yet such interventions can inadvertently increase BP variability. Our results suggest that excessive BP fluctuations, especially in hypertensive patients, may represent an underrecognized modifiable risk factor for AKI in the neurocritical care setting [13]. Elevated SBP in tICH patients may not necessarily indicate pre-existing hypertension but rather a compensatory or therapeutically induced response to maintain CPP under elevated ICP. Thus, the subgroup with mean SBP ≥140 mmHg may represent a CPP-optimized rather than chronically hypertensive population. Even so, excessive systolic BP variability was associated with a higher risk of AKI, suggesting that abrupt BP fluctuations may be harmful regardless of their origin.
Several clinical implications arise from this work. First, BPV should be monitored as a routine component of ICU management in tICH patients, especially those at high risk of AKI. Second, strategies that reduce BP fluctuations, such as more consistent use of vasoactive agents, avoidance of aggressive volume shifts, or individualized BP targets, may help mitigate AKI risk. Third, incorporating BPV into predictive models could improve early identification of high-risk patients and guide resource allocation, such as earlier nephrology consultation or renal protective strategies. In practice, continuous arterial pressure monitoring and automated BPV assessment may facilitate early detection of hemodynamic instability. Careful titration of vasoactive agents, adequate sedation, and gradual BP adjustment could help minimize harmful fluctuations in neurocritical care.
There are several limitations to this study. First, observational analysis based on the MIMIC-IV database has limitations in generalization to other institutions or healthcare systems. Second, MIMIC data has the potential for problems with missing values and potential inaccuracies in the data. Third, baseline serum creatinine was defined as the lowest value within 7 days before ICU admission.
This approach might have led to an overestimation of early AKI, particularly in patients with preexisting renal dysfunction or hemodynamic instability on admission. Finally, we evaluated BPV only during the first 24 hours after admission to the ICU, not long-term trends. Future prospective studies are needed to confirm these findings and to explore whether targeted interventions to reduce BPV can translate into improved renal outcomes in tICH patients.
CONCLUSION
Our study highlights systolic BPV as independent predictors of AKI in patients with tICH, a subtype of TBI. These findings add to the growing body of evidence that BPV is a clinically meaningful hemodynamic marker and should be considered in the management of critically ill populations. Given the high burden of AKI in tICH patients, integrating BPV into routine monitoring and clinical decision-making could help improve outcomes in this vulnerable group.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation 
Supplement1
Print 
