You are currently viewing a new version of our website. To view the old version click .
MDPIMDPI
Search all articles
Life
Download PDF
  • Article
  • Open Access

12 December 2025

Fluid Creep as an Independent Predictor of Fluid Overload and Mortality in Critically Ill Patients: A Cohort Study

,
,
,
and
1
Postgraduate Program “Emergency and Intensive Care in Children, Adolescents and Young Adults”, School of Medicine, University of Crete, 70013 Heraklion, Greece
2
Department of Intensive Care Medicine, University Hospital of Heraklion, 71500 Heraklion, Greece
3
Pediatric Intensive Care Unit, University Hospital of Heraklion, 71500 Heraklion, Greece
*
Author to whom correspondence should be addressed.
Life2025, 15(12), 1900;https://doi.org/10.3390/life15121900
This article belongs to the Special Issue Innovations in Critical Care and Anesthesiology
Version Notes
Order Reprints

Abstract

Background: Fluid overload (FO) is a frequent ICU complication and an important predictor of adverse outcomes. While classically attributed to resuscitative fluids, recent data emphasize the contribution of non-therapeutic “fluid creep” from medication diluents and carrier infusions. This study examined associations between fluid creep, FO, acute kidney injury (AKI), and mortality, and explored the predictive value of the modified Renal Angina Index (mRAI) for AKI risk stratification and FO; Methods: A retrospective cohort of 250 critically ill adults (ICU stay ≥72 h) admitted to a mixed medical–surgical ICU between May 2021 and November 2024 was analyzed. All fluids administered during the first 72 h were categorized and indexed to ideal body weight. Fluid creep included drug diluents, carriers, and flushes. FO% was calculated as [(Cumulative Fluid Balance)/IBW] × 100; Results: Fluid creep was higher in non-survivors (5183 ± 2541 vs. 4354 ± 2171 mL; p = 0.008) and correlated with FO, cumulative balance, and total input (r = 0.41 to 0.43; p < 0.001). Creep and FO independently predicted ICU mortality. Abnormal mRAI scores were associated with FO and early AKI; Conclusions: Fluid creep and FO were independent mortality predictors. Routine monitoring and minimization of creep, along with structured de-resuscitation protocols, may improve outcomes in critically ill adults.

1. Introduction

Intravenous fluid therapy is both lifesaving and potentially harmful in critical illness. It is one of the most frequently administered hospital interventions, particularly in emergency and intensive care settings. While essential for restoring hemodynamic stability, excessive administration predisposes patients to fluid overload (FO)—commonly defined as a >10% increase in body weight—which is consistently associated with impaired oxygenation, organ dysfunction, prolonged ICU stay, and increased mortality [1,2,3].
Fluid overload contributes to tissue edema and organ dysfunction through multiple mechanisms, causes morphological organ damage, and may even be fatal. Cerebral edema may exacerbate delirium and impair cognition; myocardial edema can reduce contractility and impair electrical conduction; pulmonary edema limits gas exchange and decreases compliance, increasing ventilatory support needs; and renal and gastrointestinal edema may impair filtration, tubular reabsorption, and barrier integrity [4,5]. Collectively, these effects underscore FO as a modifiable contributor to ICU morbidity and mortality.
Most studies have focused on total fluid balance without distinguishing between categories of fluid exposure [6]. Increasing evidence highlights the contribution of fluid creep—unintended fluids from drug diluents, carrier solutions, and catheter flushes [7]. Although FO is a well-established risk factor for acute kidney injury (AKI) [8] and prolonged mechanical ventilation [7,9,10], the independent role of fluid creep remains insufficiently defined.
Early identification of patients at risk for AKI is equally important to guide fluid management. The modified Renal Angina Index (mRAI), adapted from pediatric populations, integrates baseline risk factors and early markers of kidney injury into a composite score, improving prediction of severe AKI [11]. Coupling such risk stratification with structured fluid stewardship—the systematic monitoring, rationalization, and reduction of unnecessary fluid exposure—may be critical to mitigating iatrogenic harm from hidden fluid burden.
Pediatric and experimental studies suggest that fluid creep may constitute 30–60% of daily intake and substantially increase sodium and chloride burden [12,13]. In adults, non-resuscitative fluids can account for up to 42% of intake during the first three Intensive Care Unit (ICU) days, and minimizing their use has been associated with reduced FO [14,15]. One large adult study found fluid creep and maintenance fluids to be the predominant contributors to early fluid intake, particularly among mechanically ventilated patients [16]. Despite these signals, the prognostic relevance of fluid creep, independent of total fluid balance, has not been systematically evaluated [17,18].
To address these gaps, we investigated early fluid administration patterns and their associations with fluid overload FO and mortality in critically ill adults, with particular focus on fluid creep as a distinct, potentially modifiable contributor to hidden fluid burden. Exploratory analyses also assessed the correlations of fluid creep with prolonged mechanical ventilation, vasoactive support, severity of illness, and ICU length of stay and the predictive value of the modified Renal Angina Index (mRAI) for early AKI risk stratification and FO.

2. Materials and Methods

2.1. Study Design and Setting

This retrospective observational cohort study was conducted in the mixed medical–surgical ICU of the University General Hospital of Heraklion, a tertiary academic center in Greece. All adult patients admitted between 1 May 2021 and 30 November 2024, were screened for eligibility. The study was approved by the institutional review board (ID 17747/26-05-2023; ID 21549/18-10-2024/extension), with informed consent waived owing to its retrospective design. All procedures adhered to the Declaration of Helsinki and institutional research guidelines.

2.2. Patient Selection

Patients aged 18–80 years who remained in the ICU for at least 72 h were eligible for inclusion. Exclusion criteria comprised pre-existing chronic kidney disease, severe heart failure requiring fluid restriction, end-of-life palliative care, and incomplete critical data (i.e., missing essential clinical or laboratory parameters).

2.3. Data Collection

Data were extracted from electronic health records, including: Demographics: age, sex, body weight, and height (used to calculate ideal body weight [IBW] for normalization across sex and BMI categories). Admission characteristics: primary diagnosis, pre-admission data, comorbidities, admission type (medical/surgical). Severity indices: Acute Physiology and Chronic Health Evaluation II (APACHE II), Sequential Organ Failure Assessment (SOFA), modified Renal Angina Index (mRAI), Kidney Disease: Improving Global Outcomes (KDIGO) stage. Laboratory/clinical data: admission and daily measurements for 72 h. Interventions: mechanical ventilation, vasopressors, CRRT, nephrotoxic/contrast exposure.

2.4. Fluid Data

All fluids administered during the first 72 h were quantified and indexed to IBW. Categories included fluid creep (drug diluents, carrier infusions, flushes), maintenance, resuscitation (≥10 mL/kg crystalloids), replacement, nutrition, and blood products. Solutions were classified as balanced crystalloids or 0.9% sodium chloride (NS). Medication concentrations were maximized per protocol to reduce fluid burden. The handling of continuously administered medications was standardized by means of AI-driven algorithms embedded within the electronic database, which calculated the type and minimum volume of diluents required according to the prescribed dose and infusion rate. All fluid data were automatically extracted from the electronic medication administration records (recorded in real time during infusions) and were not calculated manually. Resuscitation fluids, maintenance fluids, blood products, and nutritional support were administered in accordance with current guidelines specific to each critical illness [19,20,21]. Fluid balance was calculated daily and cumulatively as Fluid Balance (L) = Total Input (L) − Total Output (L). Fluid overload percentage (FO%) was defined using the formula [22] FO% = [(Cumulative Fluid Balance in L)/IBW in kg] × 100. FO > 10% of IBW defined clinically significant FO, and FO > 15% defined severe FO.

2.5. Acute Kidney Injury and mRAI

AKI was defined using KDIGO criteria, with severe AKI defined as stage ≥2 at admission or on day 3. Renal function was monitored daily (serum creatinine, urea, estimated creatinine clearance). The mRAI was calculated within 12 h of admission by multiplying risk and injury domains [23,24]. A score ≥6 indicated high risk for severe AKI. A schematic of the scoring system is provided in Figure S1.

2.6. Statistical Analysis

Analyses were performed with SPSS v30 (IBM, Armonk, NY, USA). Continuous variables were tested for normality (Shapiro–Wilk) and expressed as mean ± SD or median (IQR). Categorical variables were reported as n (%). Between-group comparisons used ANOVA or Kruskal–Wallis tests for continuous variables and chi-square or Fisher’s exact tests for categorical variables. Correlations between fluid balance, fluid overload (FO), and individual fluid categories with illness severity, mRAI, and other clinical variables were assessed using Pearson’s correlation coefficient. Multivariable logistic regression (backward stepwise) was used to identify independent predictors of ICU mortality, FO > 10%, and severe AKI. All clinically relevant covariates (APACHE II, SOFA, KDIGO day-3 stage ≥ 2, mRAI, mechanical ventilation, vasopressor use, age, sex, and fluid variables) were included as candidate predictors. Univariate binary logistic regressions identified individual predictors. Variables with p < 0.10 or clinical relevance were subsequently included in a multivariable model. A backward stepwise likelihood-ratio (BSTEP-LR) model was used to derive the most parsimonious set of predictors. Collinearity was assessed via variance inflation factors (VIF). Model discrimination and calibration were evaluated using the area under the ROC curve (AUC), Brier score, Cox & Snell R2, Nagelkerke R2, and the Hosmer–Lemeshow goodness-of-fit test. Predictive performance of fluid creep, input, FO%, and fluid balance for ICU mortality was evaluated using ROC analysis. Significance was set at p < 0.05.

3. Results

3.1. Demographic and Clinical Characteristics

Between May 2021 and November 2024, 2000 patients were admitted to the ICU. After exclusions, 250 patients had complete datasets (Figure S2). Of these, 160 (64%) were male and 90 (36%) females, with a mean age of 64.8 ± 17 years, body weight 80.1 ± 17 kg, BMI 28.2 ± 5.6, and admission APACHE II score 21.9 ± 7.8. Mean ICU and hospital stays were 15.1 ± 16 and 35.6 ± 31 days, respectively (Table 1).
Table 1. Demographic and clinical characteristics of the study population stratified by survival status.
Non-survivors had lower body weight (p = 0.006), higher SOFA scores (p = 0.032), lower GCS (p = 0.049), and were more often admitted for medical indications (72.2% vs. 56.7%, p = 0.020). ICU-acquired infections and infections at admission were also more common in non-survivors (both p < 0.05). ICU mortality was 20%; mortality among enrolled patients was 31.6%, and hospital mortality was 47.8%. Non-survivors required more vasoactive days (p < 0.001) and had higher rates of mechanical ventilation and vasopressor use (both p < 0.001).

3.2. Laboratory Findings

On day 3, non-survivors had lower pH (p = 0.042), bicarbonate (p = 0.009), base excess (p = 0.040), and PaO2 (p = 0.025), with persistently higher lactate from day 2 onward (p < 0.001) (Table S1). Longitudinally, sodium, chloride, and potassium increased (all p < 0.001), while glucose and lactate declined in survivors (p < 0.001). Arterial blood gases improved from day 1 except for PaO2, which decreased (p < 0.001). Sodium and chloride were not associated with fluid creep, FO%, input, creatinine clearance, or outcomes.

3.3. Acute Kidney Injury

At admission, 24.5% had KDIGO stage 2 and 26.6% stage 3 AKI. By day 3, severe AKI (KDIGO ≥ 2) decreased among survivors (14.6%) but rose in non-survivors (27.7%, p = 0.011). By day 7, the gap widened (12.8% vs. 32.0%, p < 0.001) (Table 2). Creatinine clearance improved in survivors (p < 0.001) but not in non-survivors. CRRT use was higher among non-survivors (34.2% vs. 14.8%, p < 0.001). Nephrotoxic drug and diuretic exposure did not differ.
Table 2. Distribution of acute kidney injury (AKI) markers, treatments, and nephrotoxic exposures by outcome group.
Abnormal mRAI (≥6) within 12 h was more common in ICU non-survivors (32.9% vs. 21.1%, p = 0.044) and hospital non-survivors (31.9% vs. 18.5%, p = 0.014). mRAI independently predicted severe AKI in medical (OR 0.93, 95% CI 0.87–0.99, p = 0.032) and emergency patients (OR 0.94, 95% CI 0.88–0.99, p = 0.023).

3.4. Categories of Administered Fluids

3.4.1. Pre-ICU

Among 73 patients with data, mean intake was 1589 ± 1351 mL crystalloids (25 ± 22 mL/kg IBW) and 957 ± 123 mL blood products/albumin (15 ± 14 mL/kg IBW), with no subgroup differences.

3.4.2. ICU Days 1–3

During the first three ICU days, fluid creep and maintenance fluids each contributed >40% of total intake, while resuscitation and replacement accounted for ~14% and nutrition/blood products ~5% (Table S2, Figure S3). Indexed to IBW, boluses, resuscitation fluids, albumin, and blood products decreased over time, whereas maintenance, creep, and nutrition increased (Table S3).
Cumulative fluid creep was higher in non-survivors (5183 ± 2541 vs. 4354 ± 2171 mL, p = 0.008), a difference persisting when nutrition and blood products were included (“fluid creep plus”: 6342 ± 3044 vs. 5158 ± 2565 mL, p = 0.002; “fluid creep plus indexed to IBW”: 100.7 ± 44.6 vs. 78.3 ± 37.9 mL/IBW, p < 0.001) (Table S4). Among input categories, antibiotic diluents (day 2), vasoactive infusions (days 2–3), blood products (day 3), and enteral nutrition (day 1) were significantly higher in non-survivors (all p < 0.05) (Figure 1).
Figure 1. Distribution of daily fluid input by category through ICU day 3, indexed to ideal body weight (IBW) and stratified by survival status. Non-survivors received higher proportions of fluid creep, blood products (day 3), and enteral nutrition (day 1). * p < 0.05.
Other intake categories did not differ; therefore, overall input was greater in non-survivors (p = 0.002). Daily and cumulative output volumes were lower in non-survivors, though differences did not reach significance (Figure 2).
Figure 2. Distribution of daily fluid output by category through ICU day 3, indexed to ideal body weight (IBW) and stratified by survival status.

3.5. Fluid Types

Before ICU, PlasmaLyte and Ringer’s Lactate predominated. During ICU stay, boluses were mostly NS (56–62% day 1, 80–90% days 2–3). Maintenance fluids were largely Ringer’s Lactate with 5–20% D/W. Creep consisted mainly of NS, D/W 5%, and sterile water, used as diluents. Overall composition (~50% balanced crystalloids, 25% NS, 25% D/W/nutrition/blood products) did not differ between survivors and non-survivors.

3.6. Fluid Overload

By day 3, clinically significant FO (>10%) and severe FO (>15%) were both significantly more frequent in non-survivors (34.2% and 15.2%, respectively; both p < 0.001) (Table 3). Non-survivors also had higher cumulative fluid creep, input, and positive balances.
Table 3. Cumulative input, output, fluid balance, and fluid overload by ICU Day 3, stratified by survival status.
Cumulative fluid creep correlated with FO (r = 0.41, p < 0.001), balance (r = 0.41, p < 0.001), and input (r = 0.43, p < 0.001), and more weakly with mRAI score (r = 0.17, p = 0.008) and output (r = 0.17, p = 0.006) (Figure 3). It was also associated with longer ICU stay, prolonged mechanical ventilation, and increased vasoactive days (all p < 0.001). Cumulative fluid balance was positively correlated with cumulative resuscitation (r = 0.21, p < 0.001) and maintenance fluids (r = 0.20, p = 0.002) on day 3, as well as with the mRAI score (r = 0.257, p < 0.001) and the markers of illness severity including APACHE II (r = 0.30, p < 0.001) and SOFA score (r = 0.15, p = 0.016). Among all fluid categories, only cumulative resuscitation fluid demonstrated a significant association with SOFA score (r = 0.205, p = 0.002). No meaningful correlations were observed between fluid exposure variables and ICU length of stay, duration of mechanical ventilation, or duration of vasopressor support.
Figure 3. Correlation of cumulative fluid creep with (A) total input and (B) fluid balance through ICU Day 3, both indexed to IBW. Fluid creep correlated strongly with intake and fluid balance and more weakly with output, underscoring its role in fluid accumulation.
In logistic regression, only APACHE II (OR 1.14, 95% CI 1.06–1.22, p < 0.001), mRAI (OR 1.08, 95% CI 1.03–1.14, p = 0.004), cumulative fluid creep (OR 1.04, 95% CI 1.02–1.06, p < 0.001), and maintenance (OR 1.03, 95% CI 1.01–1.05, p = 0.005), but not resuscitation fluid volume, independently predicted FO > 10%. For FO > 15%, the same predictors remained significant except for maintenance. The model demonstrated very good explanatory performance (Cox & Snell R2 = 0.293; Nagelkerke R2 = 0.469) and excellent calibration (Hosmer–Lemeshow χ2 = 3.33–3.35; p ≈ 0.91). Classification accuracy was ~85% overall (96% for non-FO patients; 36% for FO > 10%), and no problematic multicollinearity was detected (all VIF < 5).

3.7. Outcomes

On univariate logistic regression, cumulative fluid creep on day 3 indexed to IBW, total FO% on day 3, total fluid input, KDIGO day-3 stage ≥ 2, and SOFA score were significantly associated with ICU mortality (Table S5). When all clinically relevant variables were entered simultaneously in a multivariable forced-entry model (Table S6), only cumulative fluid creep (aOR 1.017, 95% CI 1.003–1.030; p = 0.017) and FO (aOR 1.062, 95% CI 1.001–1.126; p = 0.046) remained independently associated with mortality. Inclusion of MV and vasopressor support attenuated the significance of other predictors, consistent with their interdependence with fluid balance and illness severity. Model performance was acceptable (AUC 0.71; Brier 0.19). In the primary backward stepwise likelihood-ratio model (BSTEP-LR; Table S7), the same two variables, fluid creep (OR 1.02, 95% CI 1.01–1.03, p = 0.012) and FO (OR 1.06, 95% CI 1.007–1.13, p = 0.028), were retained as independent predictors with superior discrimination (AUC 0.95; Brier 0.07), moderate explanatory power (Nagelkerke R2 = 0.229; Cox & Snell R2 = 0.164), and good calibration (Hosmer–Lemeshow p = 0.121). Multicollinearity was assessed via variance inflation factors (all VIF < 5).
ROC analysis confirmed predictive value for input (AUC 0.674, 95% CI 0.594–0.754, p < 0.001), FO% (0.636, 95% CI 0.547–0.725, p = 0.003), fluid balance (0.636, 95% CI 0.547–0.725, p = 0.003), maintenance fluids (0.625), 95% CI 0.539–0.710, p = 0.004, and creep (0.623, 95% CI 0.539–0.707, p = 0.004) (Figure 4). Detailed cutoff and sensitivity/specificity data are provided in the Supplement (Table S8).
Figure 4. Receiver operating characteristic (ROC) curves for cumulative fluid indices indexed to IBW in predicting ICU mortality. Total input, FO%, fluid balance, maintenance fluids, and cumulative fluid creep showed modest but significant discriminatory ability.

4. Discussion

In this retrospective adult ICU cohort, early fluid creep—unintended fluids from drug diluents, carrier solutions, and catheter flushes—was independently associated with FO and ICU mortality, even after adjusting for illness severity, conventional fluid categories, and early AKI markers. Fluid creep also correlated with prolonged mechanical ventilation, vasoactive support, and ICU length of stay, supporting a pathophysiologic role of hidden fluid burden in promoting interstitial edema and organ dysfunction.
Fluid accumulation contributes to organ dysfunction through multiple mechanisms and can lead to structural tissue injury and death, reinforcing FO as a modifiable driver of morbidity in the ICU. Cerebral edema from excess fluid may exacerbate delirium and impair cognition; myocardial edema can reduce contractility and precipitate conduction disturbances and diastolic dysfunction; pulmonary edema compromises gas exchange, reduces compliance, and increases ventilatory support needs; and renal or gastrointestinal edema may impair filtration, tubular reabsorption, nutrient absorption, and mucosal barrier integrity. These pathophysiological effects support the observed associations between FO and adverse outcomes and underscore the importance of targeted fluid stewardship [4,5].
These findings align with prior reports showing that maintenance fluids and fluid creep often exceed resuscitative boluses in their contribution to daily fluid, sodium, and chloride loads. In pediatric cohorts, higher intake—rather than reduced output—was the main driver of FO, which was associated with longer ICU stay, higher AKI incidence, fewer ventilator-free days, and increased mortality [2,25,26,27]. In a large mixed ICU cohort, creep accounted for ~33% of daily intake—fivefold greater than resuscitation fluids—and was a major source of electrolyte burden [13,28]. Likewise, among patients receiving respiratory support, fluid creep represented ~25% of intake within 24 h and persisted in those with severe hypoxemia [16]. Our results extend these observations by demonstrating that fluid creep is not only common but also prognostically relevant in critically ill adults.
In our analysis, cumulative fluid creep independently predicted clinically significant FO (>10%) and severe FO (>15%), both of which were associated with mortality. These findings reinforce that fluid creep—rather than resuscitation fluids—is the primary driver of early fluid accumulation leading to clinically significant FO in this mixed-ICU cohort. Notably, both fluid creep and FO independently predicted ICU death. These findings highlight fluid creep as a modifiable early exposure, reinforcing the rationale for targeted interventions such as optimizing drug concentrations, minimizing carrier volumes, and preferentially using balanced diluents [15]. The SALT-ED and SMART trials support this biological plausibility, demonstrating reduced kidney injury with balanced fluids compared with saline [29]. Moreover, structured fluid stewardship—progressing through resuscitation, optimization, stabilization, and de-escalation phases—provides a framework to individualize volume targets, prevent FO, and improve outcomes [30].
Consistent with broader ICU evidence, FO in our study was associated with impaired creatinine clearance and increased use of CRRT among non-survivors [31,32]. Notably, abnormal mRAI scores within the first 12 h were more frequent in non-survivors and independently predicted severe AKI, particularly in medical and emergency subgroups. These results align with previous validation studies showing that the mRAI surpasses serum creatinine changes alone in identifying patients at risk for AKI progression [11,31,32]. Although initially validated in pediatric populations [33,34], emerging data support its adaptation for adult and cardiac ICU cohorts, demonstrating comparable predictive accuracy for AKI and other adverse renal outcomes. In critically ill adults, mRAI scores modified by substituting the transplantation variable with diabetes or sepsis, when determined within the first 24 h of ICU admission, outperformed early serum creatinine changes in predicting the development of AKI Stage ≥ 2 during days 2–7 of ICU stay [11]. Furthermore, the RAI has been identified as a strong predictor not only of severe AKI but also of long-term adverse outcomes, enhancing 12-month risk stratification among cardiac ICU patients [35]. While exploratory, our findings support the potential integration of the mRAI into early fluid stewardship protocols to identify high-risk patients for whom minimizing fluid creep and maintenance fluid exposure may be most beneficial [23,35].
Fluid composition also matters. A meta-analysis of five ICU studies (n = 1105) found that balanced crystalloids, when used for creep and maintenance, significantly reduced daily sodium burden compared with saline [36]. Similarly, a prospective before–after study showed that substituting 5% glucose for maintenance and diluent fluids halved daily sodium load and lowered daily fluid balance [17]. In our cohort, predominant use of balanced crystalloids, maintained acid–base homeostasis and preserved normal sodium and chloride levels, with no association between electrolyte trends and outcomes. In line with this, a randomized controlled trial found higher chloride concentrations with saline compared with balanced crystalloids (111 vs. 108 mmol/L) but no difference in AKI or mortality [37]. A recent systematic review, however, suggested that balanced crystalloids probably reduce 90-day mortality compared with saline, underscoring the importance of fluid type even in hidden sources such as diluents [38].
From a clinical perspective, our results argue for explicit accounting of fluid creep in early ICU care and its incorporation into fluid stewardship bundles. Practical measures include daily reporting of creep volumes, use of concentrated infusions, balanced diluents when compatible, and integration with the “four Ds” (drug, dosing, duration, de-escalation) [39,40]. Equally important, structured deresuscitation protocols—emphasizing active fluid removal with diuretics or timely initiation of renal replacement therapy—are a complementary step to prevent progression from clinically significant to severe FO and thereby reduce morbidity and mortality [41,42,43]. Although our ROC analyses showed only modest discrimination (AUC 0.62–0.67), creep- and FO-related variables may still function as operational triggers to reassess volume status, initiate deresuscitation, or minimize creep exposure, rather than as strict prognostic thresholds.
This study systematically disaggregated fluid categories indexed to IBW, evaluated early exposures, and linked creep to patient-centered outcomes while adjusting for illness severity and early AKI indices [13]. Limitations include its retrospective, single-center design, potential residual confounding from illness severity and medication burden, and restricted ability to evaluate the impact of fluid type. We acknowledge that detailed data on pre-ICU fluid administration were unavailable, which may have influenced early fluid balance through perioperative fluid delivery, variable equilibration of the interstitial fluid compartment following excess pre-admission intake, fluid redistribution, and potential disease-specific effects on fluid handling [5]. Finally, mRAI analyses were exploratory, not powered for mortality prediction, and ROC performance was modest.
Given the rapidly expanding role of artificial intelligence in critical care, emerging machine-learning approaches, embedded within human-centered clinical systems and fostered through transdisciplinary collaboration, may leverage high-dimensional data on patient characteristics, hemodynamics, and fluid exposure to personalize fluid therapy and enable earlier prediction of fluid overload risk. Such tools have the potential to enhance situational awareness, support timely decision-making, and ultimately improve patient outcomes [44].

5. Conclusions

Fluid creep is an underrecognized but modifiable driver of fluid overload and ICU mortality. Incorporating its measurement and reduction into fluid stewardship programs offers a pragmatic strategy to minimize unintended harm. Prospective trials—testing concentration protocols, balanced diluents, and automated monitoring—are warranted to determine whether reducing hidden fluid exposure can improve outcomes in critically ill patients.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/life15121900/s1, Figure S1: Modified Renal Angina Index (mRAI). The mRAI was calculated 12 h after ICU admission as the product of risk and injury domains. The creatinine score was defined as the difference between serum creatinine at ICU admission and the most recent value within the preceding 3 days. An mRAI score ≥6 indicated high risk for acute kidney injury (AKI); Figure S2: Flowchart of patient screening, exclusions, and final study cohort, according to STROBE recommendations; Table S1: Laboratory Tests During the First Three Days of ICU Stay; Table S2: Distribution of daily fluid intake categories stratified by outcome category; Table S3: Distribution of daily fluid intake indexed by ideal body weight (IBW), stratified by outcome category; Figure S3: Distribution of cumulative fluid input categories by ICU day 3, indexed to IBW. Fluid creep and maintenance fluids each accounted for >40% of total intake, whereas resuscitation, replacement, and nutrition/blood products contributed smaller proportions; Table S4: Distribution of cumulative fluid categories by outcome category; Supplementary Methods: Supplementary logistic regression analyses were conducted to assess the robustness of associations between fluid variables and ICU mortality. Univariate binary logistic regressions identified individual predictors (Table S5). Variables with p < 0.10 or clinical relevance were subsequently included in a multivariable forced-entry model (Table S6). A backward stepwise likelihood-ratio (BSTEP-LR) model was used to derive the most parsimonious set of predictors (Table S7). Collinearity was assessed via variance inflation factors (VIF > 5 = moderate; VIF > 10 = high). Model discrimination and calibration were evaluated using the area under the ROC curve (AUC), Brier score, Cox & Snell R2, Nagelkerke R2, and the Hosmer–Lemeshow goodness-of-fit test. Analyses were performed in SPSS v30 (IBM Corp.) and verified in Python 3.12 (statsmodels v0.14); Table S5: Univariate Logistic Regression for ICU Mortality. Selected variables significantly associated with ICU mortality in univariate logistic regression; Table S6: Multivariable Logistic Regression (Forced-Entry Model); Table S7: Backward Stepwise (Likelihood-Ratio) Logistic Regression; Table S8: Predictive ability of cumulative fluid variables by day 3, indexed to ideal body weight (IBW) for ICU mortality, based on ROC analysis.

Author Contributions

Conceptualization, G.B., S.I. and E.K.; methodology, G.B. and S.I.; software, G.B. and S.I.; validation, G.B. and S.I.; formal analysis, T.A. and J.V.; investigation, T.A. and J.V.; resources, T.A. and J.V.; data curation, G.B. and S.I.; writing—original draft preparation, G.B.; writing—review and editing, G.B., T.A., J.V. and S.I.; visualization, E.K.; supervision, G.B. and S.I. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of the University Hospital of Heraklion, Crete (ID 17747/26 May 2023; 21549/18 October 2024).

Data Availability Statement

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Acknowledgments

The Special Account for Research Funds of the University of Crete (KA 3650) supported publication fees.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
AKIAcute Kidney Injury
FOFluid Overload
ICUIntensive Care Unit
mRAImodified Renal Angina Index
IBMIdeal Body Weight

References

  1. Alobaidi, R.; Morgan, C.; Basu, R.K.; Stenson, E.; Featherstone, R.; Majumdar, S.R.; Bagshaw, S.M. Association Between Fluid Balance and Outcomes in Critically Ill Children: A Systematic Review and Meta-Analysis. JAMA Pediatr. 2018, 172, 257–268. [Google Scholar] [CrossRef] [PubMed]
  2. Lintz, V.C.; Vieira, R.A.; de Lima Carioca, F.; de Siqueira Ferraz, I.; Silva, H.M.; Ventura, A.M.C.; de Souza, D.C.; Brandão, M.B.; Nogueira, R.J.N.; de Souza, T.H. Fluid Accumulation in Critically Ill Children: A Systematic Review and Meta-Analysis. EClinicalMedicine 2024, 74, 102714. [Google Scholar] [CrossRef] [PubMed]
  3. Renaudier, M.; Lascarrou, J.-B.; Chelly, J.; Lesieur, O.; Bourenne, J.; Jaubert, P.; Paul, M.; Muller, G.; Leprovost, P.; Klein, T.; et al. Fluid Balance and Outcome in Cardiac Arrest Patients Admitted to Intensive Care Unit. Crit. Care 2025, 29, 152. [Google Scholar] [CrossRef] [PubMed]
  4. Messina, A.; Bakker, J.; Chew, M.; De Backer, D.; Hamzaoui, O.; Hernandez, G.; Myatra, S.N.; Monnet, X.; Ostermann, M.; Pinsky, M.; et al. Pathophysiology of Fluid Administration in Critically Ill Patients. Intensive Care Med. Exp. 2022, 10, 46. [Google Scholar] [CrossRef]
  5. Hahn, R.G. Where Does the Fluid Go? Ann. Intensive Care 2025, 15, 156. [Google Scholar] [CrossRef]
  6. Pfortmueller, C.A.; Dabrowski, W.; Wise, R.; van Regenmortel, N.; Malbrain, M.L.N.G. Fluid Accumulation Syndrome in Sepsis and Septic Shock: Pathophysiology, Relevance and Treatment-a Comprehensive Review. Ann. Intensive Care 2024, 14, 115. [Google Scholar] [CrossRef]
  7. Sharif, S.; Flindall, H.; Basmaji, J.; Ablordeppey, E.; Díaz-Gómez, J.L.; Lanspa, M.; Nikravan, S.; Piticaru, J.; Lewis, K. Critical Care Ultrasonography for Volume Management: A Systematic Review, Meta-Analysis, and Trial Sequential Analysis of Randomized Trials. Crit. Care Explor. 2025, 7, e1261. [Google Scholar] [CrossRef]
  8. Woodward, C.W.; Lambert, J.; Ortiz-Soriano, V.; Li, Y.; Ruiz-Conejo, M.; Bissell, B.D.; Kelly, A.; Adams, P.; Yessayan, L.; Morris, P.E.; et al. Fluid Overload Associates with Major Adverse Kidney Events in Critically Ill Patients with Acute Kidney Injury Requiring Continuous Renal Replacement Therapy. Crit. Care Med. 2019, 47, e753–e760. [Google Scholar] [CrossRef]
  9. Vignon, P.; Evrard, B.; Asfar, P.; Busana, M.; Calfee, C.S.; Coppola, S.; Demiselle, J.; Geri, G.; Jozwiak, M.; Martin, G.S.; et al. Fluid Administration and Monitoring in ARDS: Which Management? Intensive Care Med. 2020, 46, 2252–2264. [Google Scholar] [CrossRef]
  10. Charaya, S.; Angurana, S.K.; Nallasamy, K.; Jayashree, M. Restricted versus Usual/Liberal Maintenance Fluid Strategy in Mechanically Ventilated Children: An Open-Label Randomized Trial (ReLiSCh Trial). Indian J. Pediatr. 2025, 92, 7–14. [Google Scholar] [CrossRef]
  11. Ortiz-Soriano, V.; Kabir, S.; Claure-Del Granado, R.; Stromberg, A.; Toto, R.D.; Moe, O.W.; Goldstein, S.L.; Neyra, J.A. Assessment of a Modified Renal Angina Index for AKI Prediction in Critically Ill Adults. Nephrol. Dial. Transplant. 2022, 37, 895–903. [Google Scholar] [CrossRef]
  12. Langer, T.; D’Oria, V.; Spolidoro, G.C.I.; Chidini, G.; Scalia Catenacci, S.; Marchesi, T.; Guerrini, M.; Cislaghi, A.; Agostoni, C.; Pesenti, A.; et al. Fluid Therapy in Mechanically Ventilated Critically Ill Children: The Sodium, Chloride and Water Burden of Fluid Creep. BMC Pediatr. 2020, 20, 424. [Google Scholar] [CrossRef]
  13. Van Regenmortel, N.; Verbrugghe, W.; Roelant, E.; Van den Wyngaert, T.; Jorens, P.G. Maintenance Fluid Therapy and Fluid Creep Impose More Significant Fluid, Sodium, and Chloride Burdens than Resuscitation Fluids in Critically Ill Patients: A Retrospective Study in a Tertiary Mixed ICU Population. Intensive Care Med. 2018, 44, 409–417. [Google Scholar] [CrossRef]
  14. Messmer, A.; Pietsch, U.; Siegemund, M.; Buehler, P.; Waskowski, J.; Müller, M.; Uehlinger, D.E.; Hollinger, A.; Filipovic, M.; Berger, D.; et al. Protocolised Early De-Resuscitation in Septic Shock (REDUCE): Protocol for a Randomised Controlled Multicentre Feasibility Trial. BMJ Open 2023, 13, e074847. [Google Scholar] [CrossRef]
  15. Gamble, K.C.; Smith, S.E.; Bland, C.M.; Sikora Newsome, A.; Branan, T.N.; Hawkins, W.A. Hidden Fluids in Plain Sight: Identifying Intravenous Medication Classes as Contributors to Intensive Care Unit Fluid Intake. Hosp. Pharm. 2022, 57, 230–236. [Google Scholar] [CrossRef]
  16. Sakuraya, M.; Yoshihiro, S.; Onozuka, K.; Takaba, A.; Yasuda, H.; Shime, N.; Kotani, Y.; Kishihara, Y.; Kondo, N.; Sekine, K.; et al. A Burden of Fluid, Sodium, and Chloride Due to Intravenous Fluid Therapy in Patients with Respiratory Support: A Post-Hoc Analysis of a Multicenter Cohort Study. Ann. Intensive Care 2022, 12, 100. [Google Scholar] [CrossRef] [PubMed]
  17. Bihari, S.; Prakash, S.; Potts, S.; Matheson, E.; Bersten, A.D. Addressing the Inadvertent Sodium and Chloride Burden in Critically Ill Patients: A Prospective before-and-after Study in a Tertiary Mixed Intensive Care Unit Population. Crit. Care Resusc. 2018, 20, 285–293. [Google Scholar] [CrossRef] [PubMed]
  18. Van Regenmortel, N.; Moers, L.; Langer, T.; Roelant, E.; De Weerdt, T.; Caironi, P.; Malbrain, M.L.N.G.; Elbers, P.; Van den Wyngaert, T.; Jorens, P.G. Fluid-Induced Harm in the Hospital: Look beyond Volume and Start Considering Sodium. From Physiology towards Recommendations for Daily Practice in Hospitalized Adults. Ann. Intensive Care 2021, 11, 79. [Google Scholar] [CrossRef] [PubMed]
  19. Singer, M.; Deutschman, C.S.; Seymour, C.W.; Shankar-Hari, M.; Annane, D.; Bauer, M.; Bellomo, R.; Bernard, G.R.; Chiche, J.-D.; Coopersmith, C.M.; et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA 2016, 315, 801–810. [Google Scholar] [CrossRef]
  20. Pradelli, L.; Adolph, M.; Calder, P.C.; Deutz, N.E.; Carmona, T.G.; Michael-Titus, A.T.; Muscaritoli, M.; Singer, P. Commentary on “Guidelines for the Provision of Nutrition Support Therapy in the Adult Critically Ill Patient: The American Society for Parenteral and Enteral Nutrition”. JPEN J. Parenter. Enteral Nutr. 2022, 46, 1226–1227. [Google Scholar] [CrossRef]
  21. Nolan, J.P.; Sandroni, C.; Böttiger, B.W.; Cariou, A.; Cronberg, T.; Friberg, H.; Genbrugge, C.; Haywood, K.; Lilja, G.; Moulaert, V.R.M.; et al. European Resuscitation Council and European Society of Intensive Care Medicine Guidelines 2021: Post-Resuscitation Care. Intensive Care Med. 2021, 47, 369–421. [Google Scholar] [CrossRef]
  22. Hansen, B. Fluid Overload. Front. Vet. Sci. 2021, 8, 668688. [Google Scholar] [CrossRef]
  23. Matsuura, R.; Srisawat, N.; Claure-Del Granado, R.; Doi, K.; Yoshida, T.; Nangaku, M.; Noiri, E. Use of the Renal Angina Index in Determining Acute Kidney Injury. Kidney Int. Rep. 2018, 3, 677–683. [Google Scholar] [CrossRef] [PubMed]
  24. Basu, R.K.; Kaddourah, A.; Goldstein, S.L. AWARE Study Investigators Assessment of a Renal Angina Index for Prediction of Severe Acute Kidney Injury in Critically Ill Children: A Multicentre, Multinational, Prospective Observational Study. Lancet Child Adolesc. Health 2018, 2, 112–120. [Google Scholar] [CrossRef]
  25. Mohoric, S.; Alobaidi, R.; McGraw, T.; Joffe, A.R. The Determinants of Fluid Accumulation in Critically Ill Children: A Prospective Single-Center Cohort Study. Pediatr. Nephrol. 2025, 40, 3555–3562. [Google Scholar] [CrossRef]
  26. Rajendran, A.; Bamne, P.; Upadhyay, N.; Pandwar, U.; Shrivastava, J. Impact of Fluid Overload on Mortality Among Critically Ill Pediatric Patients: An Observational Study at a Tertiary Care Hospital in Central India. Cureus 2025, 17, e82178. [Google Scholar] [CrossRef]
  27. Rao, S.B.; Akhondi-Asl, A.; Mehta, N.; Yang, Y. Association between Early Fluid Overload and Clinical Outcomes in a Pediatric ICU. Pediatr. Res. 2025. Online ahead of print. [Google Scholar] [CrossRef] [PubMed]
  28. Molin, C.; Wichmann, S.; Schønemann-Lund, M.; Møller, M.H.; Bestle, M.H. Sodium and Chloride Disturbances in Critically Ill Adult Patients: A Protocol for a Sub-Study of the FLUID-ICU Cohort Study. Acta Anaesthesiol. Scand. 2025, 69, e70028. [Google Scholar] [CrossRef] [PubMed]
  29. Self, W.H.; Semler, M.W.; Wanderer, J.P.; Wang, L.; Byrne, D.W.; Collins, S.P.; Slovis, C.M.; Lindsell, C.J.; Ehrenfeld, J.M.; Siew, E.D.; et al. Balanced Crystalloids versus Saline in Noncritically Ill Adults. N. Engl. J. Med. 2018, 378, 819–828. [Google Scholar] [CrossRef]
  30. Messina, A.; Matronola, G.M.; Cecconi, M. Individualized Fluid Optimization and De-Escalation in Critically Ill Patients with Septic Shock. Curr. Opin. Crit. Care 2025, 31, 582–590. [Google Scholar] [CrossRef]
  31. Bouchard, J.; Soroko, S.B.; Chertow, G.M.; Himmelfarb, J.; Ikizler, T.A.; Paganini, E.P.; Mehta, R.L. Program to Improve Care in Acute Renal Disease (PICARD) Study Group Fluid Accumulation, Survival and Recovery of Kidney Function in Critically Ill Patients with Acute Kidney Injury. Kidney Int. 2009, 76, 422–427. [Google Scholar] [CrossRef]
  32. Wang, N.; Jiang, L.; Zhu, B.; Wen, Y.; Xi, X.-M. Beijing Acute Kidney Injury Trial (BAKIT) Workgroup Fluid Balance and Mortality in Critically Ill Patients with Acute Kidney Injury: A Multicenter Prospective Epidemiological Study. Crit. Care 2015, 19, 371. [Google Scholar] [CrossRef]
  33. Goldstein, S.L.; Krallman, K.A.; Roy, J.-P.; Collins, M.; Chima, R.S.; Basu, R.K.; Chawla, L.; Fei, L. Real-Time Acute Kidney Injury Risk Stratification-Biomarker Directed Fluid Management Improves Outcomes in Critically Ill Children and Young Adults. Kidney Int. Rep. 2023, 8, 2690–2700. [Google Scholar] [CrossRef]
  34. Melink, K.F.; Stanski, N.L.; Gorrell, E.; Alten, J.A.; Pettit, K.A.; Goldstein, S.L.; Zang, H.; Ollberding, N.J.; Basu, R.K.; Menon, S.; et al. Modification of the Cardiac Renal Angina Index for Predicting Adverse Kidney Events After Pediatric Cardiac Surgery. J. Am. Heart Assoc. 2025, 14, e042941. [Google Scholar] [CrossRef] [PubMed]
  35. Sakaguchi, E.; Naruse, H.; Ishihara, Y.; Hattori, H.; Yamada, A.; Kawai, H.; Muramatsu, T.; Tsuboi, Y.; Fujii, R.; Suzuki, K.; et al. Assessment of the Renal Angina Index in Patients Hospitalized in a Cardiac Intensive Care Unit. Sci. Rep. 2024, 14, 75. [Google Scholar] [CrossRef] [PubMed]
  36. Waskowski, J.; Salvato, S.M.; Müller, M.; Hofer, D.; van Regenmortel, N.; Pfortmueller, C.A. Choice of Creep or Maintenance Fluid Type and Their Impact on Total Daily ICU Sodium Burden in Critically Ill Patients: A Systematic Review and Meta-Analysis. J. Crit. Care 2023, 78, 154403. [Google Scholar] [CrossRef]
  37. Verma, B.; Luethi, N.; Cioccari, L.; Lloyd-Donald, P.; Crisman, M.; Eastwood, G.; Orford, N.; French, C.; Bellomo, R.; Martensson, J. A Multicentre Randomised Controlled Pilot Study of Fluid Resuscitation with Saline or Plasma-Lyte 148 in Critically Ill Patients. Crit. Care Resusc. 2016, 18, 205–212. [Google Scholar] [CrossRef] [PubMed]
  38. Hammond, N.E.; Zampieri, F.G.; Di Tanna, G.L.; Garside, T.; Adigbli, D.; Cavalcanti, A.B.; Machado, F.R.; Micallef, S.; Myburgh, J.; Ramanan, M.; et al. Balanced Crystalloids versus Saline in Critically Ill Adults—A Systematic Review with Meta-Analysis. NEJM Evid. 2022, 1, EVIDoa2100010. [Google Scholar] [CrossRef]
  39. Malbrain, M.L.N.G.; Van Regenmortel, N.; Saugel, B.; De Tavernier, B.; Van Gaal, P.-J.; Joannes-Boyau, O.; Teboul, J.-L.; Rice, T.W.; Mythen, M.; Monnet, X. Principles of Fluid Management and Stewardship in Septic Shock: It Is Time to Consider the Four D’s and the Four Phases of Fluid Therapy. Ann. Intensive Care 2018, 8, 66. [Google Scholar] [CrossRef]
  40. Malbrain, M.L.N.G.; Martin, G.; Ostermann, M. Everything You Need to Know about Deresuscitation. Intensive Care Med. 2022, 48, 1781–1786. [Google Scholar] [CrossRef]
  41. Gist, K.M.; Selewski, D.T.; Brinton, J.; Menon, S.; Goldstein, S.L.; Basu, R.K. Assessment of the Independent and Synergistic Effects of Fluid Overload and Acute Kidney Injury on Outcomes of Critically Ill Children. Pediatr. Crit. Care Med. 2020, 21, 170–177. [Google Scholar] [CrossRef] [PubMed]
  42. Vaara, S.T.; Korhonen, A.-M.; Kaukonen, K.-M.; Nisula, S.; Inkinen, O.; Hoppu, S.; Laurila, J.J.; Mildh, L.; Reinikainen, M.; Lund, V.; et al. Fluid Overload Is Associated with an Increased Risk for 90-Day Mortality in Critically Ill Patients with Renal Replacement Therapy: Data from the Prospective FINNAKI Study. Crit. Care 2012, 16, R197. [Google Scholar] [CrossRef] [PubMed]
  43. Liu, C.; Lu, G.; Wang, D.; Lei, Y.; Mao, Z.; Hu, P.; Hu, J.; Liu, R.; Han, D.; Zhou, F. Balanced Crystalloids versus Normal Saline for Fluid Resuscitation in Critically Ill Patients: A Systematic Review and Meta-Analysis with Trial Sequential Analysis. Am. J. Emerg. Med. 2019, 37, 2072–2078. [Google Scholar] [CrossRef] [PubMed]
  44. Hadweh, P.; Niset, A.; Salvagno, M.; Al Barajraji, M.; El Hadwe, S.; Taccone, F.S.; Barrit, S. Machine Learning and Artificial Intelligence in Intensive Care Medicine: Critical Recalibrations from Rule-Based Systems to Frontier Models. J. Clin. Med. 2025, 14, 4026. [Google Scholar] [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Journal Statistics
Article metric data becomes available approximately 24 hours after publication online.
Download PDF
Correction: Adetunji et al. DNA Methylation: A Key Regulator in Male and Female Reproductive Outcomes. Life 2025, 15, 1109
Previous
Obtaining Muramic Acid from Staphylococcus aureus: A Simple Strategy for Axenic Isolation of Tannerella forsythia
Next