Abstract
Background
Individualized management of early fluid therapy in acute pancreatitis (AP) remains challenging. Traditional studies primarily focus on static fluid balance, overlooking the heterogeneity of dynamic trajectories.
Methods
Based on a single-center retrospective cohort, 3,142 AP patients admitted within 72 h of onset were included. Daily fluid therapy-to-body weight ratio (FWR) and CumFWR (trapezoidal AUC) were calculated. Latent Class Growth Modeling (LCGM) identified dynamic FWR trajectories within 72 h. Multivariate Cox regression analyzed the relationship between FWR and in-hospital mortality risk. Three-dimensional fitted surface plot analysis evaluated interactions between CumFWR and hematocrit.
Results
LCGM successfully identified five distinct dynamic fluid therapy trajectories within 72 h, showing significant clinical differences: Low Stable Group (LSG1, n = 229, 7.3%), Low-Moderate Sustained Group (LMSG2, n = 340, 10.8%), Moderate Stable Group (MSG3, n = 2073, 66.0%), Moderate-High Fluctuating Group (MHFG4, n = 231, 7.4%), and High Sustained Group (HSG5, n = 269, 8.6%). Multivariate Cox regression revealed significantly increased in-hospital mortality risk compared to MSG3 for both MHFG4 (adjusted HR = 2.08, 95%CI 1.15–3.78) and HSG5 (adjusted HR = 2.91, 95%CI 1.77–4.79). Furthermore, each 1 standard deviation increase in CumFWR was associated with a 47% increased mortality risk (HR = 1.47, 95%CI 1.26–1.72). Three-dimensional fitted surface plot demonstrated high mortality risk between high CumFWR and abnormal hematocrit (low/high HCT), particularly prominent in the HSG5 group (p < 0.001).
Conclusion
This study reveals five distinct dynamic trajectory patterns of early fluid therapy in AP. High-load sustained or fluctuating therapy significantly increases mortality risk. Dynamic monitoring of HCT (within a 35–44% safety window) and therapy trajectories offers a potential strategy to optimize fluid management.
Supplementary Information
The online version contains supplementary material available at 10.1186/s13017-025-00664-w.
Keywords: Acute pancreatitis, Fluid therapy, Latent class growth analysis, Hematocrit, Prognosis, In-hospital mortality
Introduction
Acute pancreatitis (AP) is a common acute and critical digestive system disorder, with a global annual incidence of approximately 34 per 100,000 population [1]. Among AP patients, 20–30% progress to severe acute pancreatitis (SAP), where mortality can be as high as 30% [2]. Early fluid therapy is a cornerstone of AP treatment, aimed at correcting hypovolemia, improving microcirculatory perfusion, and preserving organ function [3–5]. However, clinical practice suggests that excessive fluid administration may lead to serious complications, such as abdominal compartment syndrome (ACS), potentially causing organ dysfunction [6–9]. Current clinical guidelines recommend weight-based fluid management, yet its application shows significant individual variability and lacks precise, dynamic adjustment strategies [10, 11].
Previous research has primarily focused on static fluid balance assessments at single time points, overlooking the heterogeneity in dynamic fluid volume trajectories over time [10, 12, 13]. Recently, longitudinal data analysis methods like Latent Class Growth Modeling (LCGM) have been applied in fields such as sepsis and kidney disease, demonstrating nonlinear associations between fluid balance trajectories and patient prognosis [14–17]. This study, based on a large observational cohort, aims to systematically explore the dynamic patterns of early fluid therapy volume (standardized by body weight as Fluid therapy Volume per Weight Ratio, FWR), its cumulative effect (Cumulative FWR, CumFWR), and their association with clinical prognosis in AP patients.
Methods
Study design and data source
This is a single-center retrospective cohort study, relying on the prospectively maintained AP-specific database from the Department of Gastroenterology, the First Affiliated Hospital of Nanchang University. This database systematically collected data from all eligible AP patients admitted consecutively between January 2005 and December 2023. The study adhered strictly to the principles of the Declaration of Helsinki and the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cohort studies [18, 19]. Approval was obtained from the Ethics Committee of The First Affiliated Hospital of Nanchang University (No. 2011001), and informed consent was waived due to data anonymization. The study population comprised adult patients diagnosed with AP according to the revised Atlanta criteria and admitted within 72 h of symptom onset [2]. Exclusion criteria were: age < 18 or > 75 years; pregnancy or lactation; pre-existing chronic organ failure (e.g., end-stage renal disease requiring dialysis, Child–Pugh class C cirrhosis); end-stage liver/kidney disease; hospital stay < 3 days; or missing key baseline data (body weight, detailed fluid volume records for the first 72 h, or outcome data). After screening, 3142 patients meeting all criteria were included in the final cohort (Fig. 1).
Fig. 1.
Patient Screening Flowchart. AP, acute pancreatitis; LOS, length of hospital stay; FWR, fluid therapy-to-body weight ratio; LSG1 (Low-stable Group, n = 229), LMSG2 (Low-moderate Sustained Group, n = 340), MSG3 (Moderate-stable Group, n = 2073), MHFG4 (Medium–high Fluctuating Group, n = 231), HSG5 (High-sustained Group, n = 269); HVT: High-volume Trajectory Group (MHFG4 + HSG5); LMVT: Low-Moderate Volume Trajectory Group (LSG1 + LMSG2 + MSG3)
Variable collection and missing data handling
Demographic and baseline characteristics include age, gender, height, weight at admission. Record the main etiology (biliary, alcoholic, hypertriglyceridemic, idiopathic, and other). Disease severity was assessed using validated scoring systems including Acute Physiology and Chronic Health Evaluation II (APACHE II), Systemic Inflammatory Response Syndrome (SIRS) criteria count (SIRS score), and revised Atlanta classification (Mild AP, Moderately Severe AP, Severe AP). Laboratory indicators at admission included hematocrit (HCT), blood urea nitrogen (BUN), creatinine, serum calcium, albumin, and triglycerides. Imaging assessments (contrast-enhanced CT [CECT] or MRI) at admission and subsequently were recorded to evaluate pancreatic necrosis extent (revised Atlanta criteria) and local complications (e.g., acute necrotic collection, walled-off necrosis). Treatment records included daily (particularly the first 72 h) total intravenous intake (mL) of crystalloids (normal saline, balanced salt solutions like lactated Ringer's), colloids (e.g., albumin), and blood products. ICU admission, ICU length of stay, and organ support therapies (mechanical ventilation, renal replacement therapy [RRT]) were also documented. The primary endpoint was all-cause in-hospital mortality. Secondary endpoints included persistent organ failure (POF; respiratory, cardiovascular, or renal failure lasting > 48 h), infected pancreatic necrosis (IPN; microbiologically or radiologically confirmed), length of hospital stay, and 28-day and 90-day all-cause mortality.
Variable definition and calculation
Daily FWR (mL/kg/day): The total daily intravenous fluid volume (Day 1, Day 2, Day 3), comprising crystalloids + colloids + blood products, divided by the patient's actual admission body weight (kg). Oral/enteral nutrition and drug vehicle volumes were generally excluded, unless the vehicle constituted the primary therapy fluid.
Cumulative FWR (CumFWR, mL/kg): Calculated as the area under the curve (AUC) of the time-intensity curve [20, 21]. Daily FWR values (vertical axis) plotted against time after admission (Days 1–3, horizontal axis) were integrated using the trapezoidal rule, quantifying total fluid therapy exposure and reflecting the early total fluid load (Fig. S1).
FWR Dynamic Trajectory Model: Latent Class Growth Modeling (LCGM) was applied to daily FWR values (Days 1–3) to identify subgroups (trajectories) with distinct temporal patterns. Model fit was assessed using the Bayesian Information Criterion (BIC), Akaike Information Criterion (AIC) (lower values indicate better fit), convergence status (Conv coefficient: 1 = converged, 2 = not converged), entropy (closer to 1 indicates clearer classification), and Average Posterior Probability (AvePP ≥ 0.7 per group for high classification reliability) [22]. Identified trajectory patterns required clear clinical significance and discrimination; groups with small proportions (< 5%) or unclear meaning were avoided. After fitting models with 2–6 trajectory classes and varying polynomial degrees, the optimal model (a 5-class solution) was selected (Table S1).
Trajectory Group: Based on LCGM patterns, clinical significance, and fluid volume levels, groups were named: Low Stable Group (LSG1), Low-Moderate Sustained Group (LMSG2), Moderate Stable Group (MSG3), Moderate-High Fluctuating Group (MHFG4), High Sustained Group (HSG5).
High Volume Trajectory Group (HVT): Defined for subsequent analyses (e.g., PSM), comprising the MHFG4 and HSG5 groups, representing sustained high-load or fluctuating high-load therapy patterns.
Low-Moderate Volume Trajectory Group (LMVT): Comprising the LSG1, LMSG2, and MSG3 groups, representing lower-volume or moderate, stable therapy patterns.
Statistical analysis
Continuous variables were assessed for normality using Shapiro–Wilk tests or histograms. Normally distributed variables are presented as mean ± standard deviation (SD), with between-group comparisons using independent samples t-tests (two groups) or one-way ANOVA (multiple groups). When homogeneity of variance was violated (Levene test), Welch's ANOVA was applied. Non-normally distributed variables are expressed as median [interquartile range, IQR], with comparisons using Mann–Whitney U tests (two groups) or Kruskal–Wallis H tests (multiple groups). Categorical variables are presented as frequencies and percentages (n [%]), compared using chi-square or Fisher's exact tests. Baseline characteristics were compared across CumFWR quartiles (Q1-Q4) (Tables S2 and S3) and FWR trajectory groups (LSG1, LMSG2, MSG3, MHFG4, HSG5) (Tables 1 and 2).
Table 1.
Baseline characteristics of included acute pancreatitis in different fluid therapy-to-body weight ratio (FWR) trajectory groups
| Variables | LSG1 (n = 229) |
LMSG2 (n = 340) |
MSG3 (n = 2073) |
MHFG4 (n = 231) |
HSG5 (n = 269) |
p |
|---|---|---|---|---|---|---|
| Male, n (%) | 184 (80.3) | 223 (65.6) | 1340 (64.6) | 144 (62.3) | 159 (59.1) | < 0.001 |
| Age (years) | 45.4 ± 13.1 | 48.7 ± 12.9 | 47.3 ± 13.3 | 46.9 ± 13.8 | 47.9 ± 14.6 | 0.056 |
| BMI (kg/m2) | 27.7 ± 3.4 | 24.2 ± 3.3 | 24.8 ± 4.1 | 24.6 ± 4.4 | 23.3 ± 3.5 | < 0.001 |
| Etiology, n (%) | < 0.001 | |||||
| Biliary | 81 (35.4) | 192 (56.5) | 929 (44.8) | 84 (36.4) | 108 (40.1) | |
| HTG | 86 (37.6) | 90 (26.5) | 697 (33.6) | 104 (45) | 104 (38.7) | |
| Alcoholism | 20 (8.7) | 23 (6.8) | 164 (7.9) | 14 (6.1) | 16 (5.9) | |
| Others | 42 (18.3) | 35 (10.3) | 283 (13.7) | 29 (12.6) | 41 (15.2) | |
| Smoking, n (%) | 64 (27.9) | 87 (25.6) | 624 (30.1) | 59 (25.5) | 77 (28.6) | 0.328 |
| Drinking, n (%) | 64 (27.9) | 75 (22.1) | 642 (31) | 88 (38.1) | 90 (33.5) | < 0.001 |
| Diabetes, n (%) | 23 (10) | 24 (7.1) | 251 (12.1) | 41 (17.7) | 48 (17.8) | < 0.001 |
| Body temperature (°C) | 37.0 ± 0.6 | 37.0 ± 0.6 | 37.0 ± 0.7 | 37.1 ± 0.8 | 37.1 ± 0.9 | 0.005 |
| Heart rate (bpm) | 89.9 ± 17.0 | 88.5 ± 14.8 | 93.6 ± 19.8 | 104.6 ± 22.5 | 104.0 ± 24.4 | < 0.001 |
| Respirations (bpm) | 20.3 ± 1.7 | 20.4 ± 2.2 | 22.2 ± 5.4 | 24.6 ± 7.0 | 24.4 ± 7.7 | < 0.001 |
|
Mean Arterial Pressure (mmHg) |
99.6 ± 14.1 | 97.5 ± 13.1 | 99.1 ± 22.0 | 98.5 ± 16.2 | 95.3 ± 15.1 | 0.03 |
| NEU (%) | 80.7 ± 10.7 | 79.2 ± 13.3 | 84.3 ± 8.8 | 84.9 ± 8.0 | 84.1 ± 9.1 | < 0.001 |
| HCT (%) | 41.8 ± 6.1 | 40.7 ± 6.6 | 40.7 ± 8.1 | 42.9 ± 8.2 | 42.6 ± 8.8 | < 0.001 |
| TBIL (umol/L) | 17.0 (10.7, 28.2) | 16.7 (10.0, 27.6) | 18.7 (12.4, 31.9) | 17.8 (11.6, 32.0) | 18.8 (10.9, 35.8) | 0.036 |
| ALB (g/L) | 41.4 ± 4.7 | 40.0 ± 4.3 | 39.1 ± 6.0 | 37.4 ± 6.8 | 38.0 ± 7.3 | < 0.001 |
| LDH (U/L) | 234.2 (159.8, 321.5) | 225.5 (127.8, 310.0) | 305.4 (218.0, 449.0) | 378.8 (246.0, 650.2) | 376.7 (260.1, 641.5) | < 0.001 |
| TG (mmol/L) | 2.0 (1.1, 6.5) | 1.8 (0.9, 4.2) | 1.9 (0.9, 7.8) | 4.8 (1.2, 13.3) | 2.1 (1.1, 12.6) | < 0.001 |
| AMY (U/L) | 191.7 (81.0, 599.0) | 240.5 (79.2, 685.8) | 400.0 (146.0, 905.0) | 469.0 (210.0, 915.0) | 525.9 (170.8, 1079.5) | < 0.001 |
| BUN (mmol/L) | 4.6 (3.7, 6.1) | 5.1 (4.0, 6.7) | 4.9 (3.6, 6.5) | 5.6 (4.2, 8.1) | 5.8 (4.3, 9.1) | < 0.001 |
| Cr (mmol/L) | 70.0 (57.1, 78.0) | 65.0 (54.8, 78.0) | 67.0 (54.0, 82.4) | 70.6 (55.6, 99.2) | 74.8 (58.9, 116.4) | < 0.001 |
| Ca (mmol/L) | 2.2 ± 0.2 | 2.2 ± 0.2 | 2.1 ± 0.3 | 2.0 ± 0.4 | 2.0 ± 0.4 | < 0.001 |
| PaO₂/FiO₂ | 410.6 ± 94.0 | 419.3 ± 100.6 | 365.5 ± 113.5 | 376.6 ± 133.7 | 376.2 ± 138.4 | < 0.001 |
| SIRS | 1.0 ± 0.9 | 1.0 ± 1.0 | 1.4 ± 1.1 | 1.9 ± 1.1 | 1.7 ± 1.2 | < 0.001 |
| APACHEII | 4.8 ± 3.4 | 4.5 ± 3.1 | 7.0 ± 4.6 | 9.9 ± 4.6 | 10.8 ± 5.3 | < 0.001 |
| ALB used, n (%) | 5 (2.2) | 25 (7.4) | 520 (25.1) | 109 (47.2) | 122 (45.4) | < 0.001 |
| LMWH used, n (%) | 29 (12.7) | 36 (10.6) | 416 (20.1) | 72 (31.2) | 77 (28.6) | < 0.001 |
| Insulin used, n (%) | 64 (27.9) | 85 (25) | 1133 (54.7) | 174 (75.3) | 210 (78.1) | < 0.001 |
| FWR day1, (mL/kg) | 24.9 ± 3.1 | 34.4 ± 3.2 | 41.6 ± 14.8 | 42.1 ± 17.3 | 68.8 ± 34.8 | < 0.001 |
| FWR day2, (mL/kg) | 25.2 ± 2.9 | 34.5 ± 3.1 | 44.3 ± 12.6 | 70.7 ± 13.8 | 76.6 ± 31.3 | < 0.001 |
| FWR day3, (mL/kg) | 25.1 ± 3.0 | 34.3 ± 3.2 | 43.2 ± 13.5 | 48.6 ± 15.5 | 75.4 ± 25.6 | < 0.001 |
| CumFWR, (mL/kg) | 50.2 ± 5.7 | 68.8 ± 6.1 | 86.7 ± 24.2 | 116.1 ± 25.9 | 148.7 ± 46.4 | < 0.001 |
Abbreviations: FWR, fluid therapy-to-body weight ratio; LSG1, low-stable group1; LMSG2, low-moderate-sustained group2; MSG3, medium-stable group3; MHFG4, medium high-fluctuating group4; HSG5, high-sustained group; BMI, Body Mass Index; NEU, neutrophil; HCT, hematocrit; TBIL, total bilirubin; ALB, albumin; TG, triglyceride; BUN, blood urea nitrogen; Cr, creatinine; Ca, calcium; SIRS, systemic inflammatory response syndrome; APACHEII, Acute Physiology and Chronic Health Evaluation II; LMWH, Low Molecular Weight Heparin
Table 2.
The clinical outcomes of included acute pancreatitis in different FWR trajectory groups
| Variables | LSG1 (n = 229) |
LMSG2 (n = 340) |
MSG3 (n = 2073) |
MHFG4 (n = 231) |
HSG5 (n = 269) |
p |
|---|---|---|---|---|---|---|
| Severity, n (%) | < 0.001 | |||||
| MAP, n (%) | 141 (61.6) | 164 (48.2) | 890 (42.9) | 64 (27.7) | 79 (29.4) | |
| MSAP, n (%) | 83 (36.2) | 146 (42.9) | 721 (34.8) | 65 (28.1) | 85 (31.6) | |
| SAP, n (%) | 5 (2.2) | 30 (8.8) | 462 (22.3) | 102 (44.2) | 105 (39) | |
| APFC, n (%) | 47 (20.5) | 113 (33.2) | 511 (24.7) | 53 (22.9) | 62 (23) | 0.003 |
| PPC, n (%) | 4 (1.7) | 4 (1.2) | 51 (2.5) | 3 (1.3) | 7 (2.6) | 0.462 |
| WON, n (%) | 3 (1.3) | 15 (4.4) | 160 (7.7) | 33 (14.3) | 61 (22.7) | < 0.001 |
| ANC, n (%) | 18 (7.9) | 43 (12.6) | 455 (21.9) | 98 (42.4) | 106 (39.4) | < 0.001 |
| IPN, n (%) | 2 (0.9) | 7 (2.1) | 80 (3.9) | 24 (10.4) | 32 (11.9) | < 0.001 |
| Respiratory failure, n (%) | 30 (13.1) | 56 (16.5) | 664 (32) | 123 (53.2) | 128 (47.6) | < 0.001 |
| Persistent respiratory failure, n (%) | 5 (2.2) | 26 (7.6) | 446 (21.5) | 98 (42.4) | 102 (37.9) | < 0.001 |
| Renal failure, n (%) | 3 (1.3) | 9 (2.6) | 153 (7.4) | 51 (22.1) | 67 (24.9) | < 0.001 |
| Persistent renal failure, n (%) | 0 (0) | 5 (1.5) | 91 (4.4) | 38 (16.5) | 55 (20.4) | < 0.001 |
| Circulatory failure, n (%) | 1 (0.4) | 5 (1.5) | 82 (4) | 32 (13.9) | 49 (18.2) | < 0.001 |
| Persistent circulatory failure, n (%) | 1 (0.4) | 4 (1.2) | 70 (3.4) | 25 (10.8) | 44 (16.4) | < 0.001 |
| Organ failure, n (%) | 33 (14.4) | 60 (17.6) | 694 (33.5) | 128 (55.4) | 134 (49.8) | < 0.001 |
| Persistent Organ Failure, n (%) | 5 (2.2) | 30 (8.8) | 462 (22.3) | 102 (44.2) | 105 (39) | < 0.001 |
| Persistent MOF, n (%) | 1 (0.4) | 4 (1.2) | 114 (5.5) | 46 (19.9) | 65 (24.2) | < 0.001 |
| LOS (days) | 7.0 (5.0, 9.0) | 8.0 (6.0, 10.2) | 9.0 (6.0, 14.0) | 12.0 (7.0, 20.0) | 11.0 (7.0, 22.0) | < 0.001 |
| Mortality, n (%) | 1 (0.4) | 2 (0.6) | 43 (2.1) | 18 (7.8) | 37 (13.8) | < 0.001 |
Abbreviations: AP, acute pancreatitis; MAP, mild acute pancreatitis; MSAP, moderately severe acute pancreatitis; SAP, severe acute pancreatitis; APFC, acute peripancreatic fluid collection; PPC, pancreatic pseudocyst; WON, walled-off necrosis; ANC, acute necrotic collection; IPN, infected pancreatic necrosis; LOS, length of hospital stay; MOF, multiple organ failure; ICU, intensive care unit
Confounding factors were identified using directed acyclic graphs and collinearity testing [23] (Supplementary Fig. S2, Table S4), including gender, age, smoking/drinking history, etiology (biliary, alcoholic, hypertriglyceridemic, other/idiopathic), body temperature, heart rate, and key laboratory indicators (admission HCT, BUN, TG). Multivariate Cox proportional hazards and logistic regression models were constructed with in-hospital mortality, SAP, persistent multiple organ failure (PMOF), and infected pancreatic necrosis (IPN) as outcomes. Daily FWR (Days 1–3) and CumFWR were analyzed as continuous variables, with adjusted hazard ratios (aHR) and odds ratios (aOR) reported per 1-SD increase. Mortality risk was analyzed using: 1) Q1 (lowest CumFWR quartile) as reference for Q2-Q4, and 2) MSG3 (moderate-risk reference) for other trajectory groups (particularly MHFG4/HSG5). Adjustment models were stratified: Model 1: Unadjusted; Model 2: Model 1 + demographics (gender, age, smoking/drinking); Model 3: Model 2 + clinical factors (etiology, temperature, heart rate); Model 4: Model 3 + laboratory indicators (HCT, BUN, TG). Primary results are from Model 4. Proportional hazards assumptions were verified via Schoenfeld residuals (Fig. S3). Missing data underwent simple (median), K-nearest neighbors (KNN), and multiple imputation; sensitivity analyses confirmed KNN robustness (Table S5, Fig. S4), justifying its selection.
Restricted cubic splines (RCS; 4 knots) were incorporated into Cox proportional hazards models to evaluate potential nonlinear relationships between CumFWR, HCT, and mortality risk, with linearity assessed via likelihood ratio tests. Kaplan–Meier curves compared cumulative survival across groups within specified time windows (90-day, ≤ 14-day, and 14–28-day periods), with log-rank tests determining statistical significance. The predicted mortality risk under different CumFWR levels and different APACHE II scores was visually displayed through three-dimensional fitted surface plots. RCS model was used to explore the nonlinear association between admission HCT and in-hospital mortality risk. The predicted mortality risk under different combinations of HCT levels and CumFWR levels was displayed through 3D fitted surface plots. Within the identified different FWR dynamic trajectory groups, the above visualization analysis of the combined effect of HCT and CumFWR (3D surface plot) was repeated to reveal the heterogeneous impact of HCT and fluid load on mortality risk under different therapy patterns.
To rigorously evaluate the independent prognostic impact of high-volume trajectories (HVT: MHFG4 + HSG5) while controlling for baseline confounding, 1:1 propensity score matching (PSM) was performed against the low-moderate volume trajectory group (LMVT: LSG1 + LMSG2 + MSG3). Propensity scores, derived from multivariable logistic regression, included covariates with standardized mean differences (SMD) > 0.1 potentially associated with both group assignment and mortality: gender, BMI, alcohol history, vital signs (temperature, respiration, heart rate), admission APACHE II/SIRS scores, and key laboratory indicators (HCT, BUN, TG). Nearest-neighbor matching without replacement was implemented within a caliper of 0.2 SDs of the logit(PS), with random HVT ordering to mitigate sequence bias. Covariate balance was confirmed by absolute SMD < 0.1 (Table S6, Fig. S5), with post-matching CumFWR distributions across SIRS/severity strata shown in Fig. S6. In the matched cohort (483 pairs), complication and mortality rates were compared using chi-square/Fisher's exact tests and Kaplan–Meier/log-rank analyses. To further verify the robustness of the PSM results, inverse probability of treatment weighting (IPTW), standardized mortality ratio weighting (SMRW), pairwise algorithmic weighting (PA), and overlap weighting (OW) were applied in the original unmatched cohort (Table S7).
In the matched cohort, we conducted subgroup analyses to assess heterogeneity in the HVT-mortality association across key baseline characteristics. Prespecified subgroups included: etiology (biliary vs. hypertriglyceridemic), gender, age (< 60 vs. ≥ 60 years), BMI (< 24 vs. ≥ 24 kg/m2), smoking status, alcohol history, hematocrit (< 44% vs. ≥ 44%), admission SIRS status (< 2 vs. ≥ 2 criteria), and hospital stay (< 28 vs. ≥ 28 days). Forest plots were used to display the results of each subgroup, and interaction terms between subgroup variables and HVT grouping were introduced for formal interaction testing (Wald test).
To evaluate the robustness of the HVT-mortality association, we performed sensitivity analyses: Temporal sensitivity: Excluding patients admitted during the first 9 years (2005–2013); Comorbidity exclusion: Removing patients with hypertension, COPD, or diabetes; Early admission subset: Restricting to patients admitted ≤ 24 h post-onset; First-episode restriction: Excluding recurrent AP cases; Disease severity subset: Including only patients with APACHE II > 8 at admission; Model augmentation: Adjusting for additional parameters (mean arterial pressure, PaO2/FiO2 ratio) and treatments (albumin, LMWH, insulin); Outcome redefinition: Assessing 28-day and 90-day mortality separately (Fig. S7). We calculated E-values to quantify the minimum unmeasured confounding strength needed to explain the HVT-mortality association (Fig. S8) [24]. All analyses were performed using R Statistical Software (Version 4.2.2, http://www.R-project.org, The R Foundation) and Free Statistics analysis platform (Version 2.2, Beijing, China) [25], with two-tailed tests (p < 0.05).
Result
Study population characteristics
The final cohort comprised 3,142 eligible AP patients (Fig. 1). Patient characteristics and outcomes were stratified by CumFWR quartiles (Fig. 2; Tables S2 and S3). Restricted cubic splines revealed a linear dose–response relationship between increasing CumFWR and rising mortality risk (Fig. 3A). Kaplan–Meier analysis demonstrated significantly lower cumulative survival in Q4 versus other quartiles at 90 days, 14 days, and 14–28 days (log-rank p < 0.01) (Fig. 3B,C).
Fig. 2.
Distribution of different clinical outcomes (A-D Death, E-H SAP, I-L IPN, M-P PMOF) by Cumulative FWR and FWR at Different Time Points (Day 1, Day 2, Day 3), Presented as Probability Density Curves and Bar Charts
Fig. 3.
Dynamic FWR trajectories and survival analysis. A Dose–effect relationship between Cumulative Fluid Therapy Volume (CumFWR) and In-hospital Mortality Risk (Restricted Cubic Spline model). B, C Kaplan–Meier survival curves for CumFWR quartile groups within 90 days (B), and within 14 days and 14–28 days (C). D Five dynamic trajectories identified by the Latent Class Growth Model: LSG1 (Low-stable Group, n = 229), LMSG2 (Low-moderate Sustained Group, n = 340), MSG3 (Moderate-stable Group, n = 2073), MHFG4 (Medium–high Fluctuating Group, n = 231), HSG5 (High-sustained Group, n = 269). E, F Kaplan–Meier curves for different trajectory groups
LCGM identified five distinct FWR trajectories within 72 h (Fig. 3D): Low Stable Group (LSG1, n = 229), Low-Moderate Sustained Group (LMSG2, n = 340), Moderate Stable Group (MSG3, n = 2073), Moderate-High Fluctuating Group (MHFG4, n = 231), and High Sustained Group (HSG5, n = 269). Baseline characteristics differed significantly across trajectory groups (Table 1), with severe complications and mortality showing gradient increases from LSG1 to HSG5 (Table 2). Moreover, among patients with different severity stratifications, APACHE II, SIRS scores, hospital stay, and ICU stay days of SAP patients showed a gradient increase along different trajectories (Fig. S9). Both MHFG4 and HSG5 showed elevated early and mid-term mortality (Fig. 3E,F).
Association between FWR and AP prognosis
Cox regression analyses further confirmed these findings. Each 1-SD increase in daily FWR (Days 1–3) and CumFWR significantly elevated in-hospital mortality risk (Model 4 adjusted HR: Day1 = 1.29, 95%CI 1.12–1.47; Day2 = 1.39, 1.22–1.58; Day3 = 1.30, 1.07–1.57; CumFWR = 1.47, 1.26–1.72; all p < 0.01). Compared to Q1 (lowest CumFWR), Q4 (highest CumFWR) showed significantly increased mortality risk (adjusted HR = 3.68, 95%CI 1.11–12.21, P = 0.03). Using MSG3 as reference, mortality risk was elevated in MHFG4 (adjusted HR = 2.08, 95%CI 1.15–3.78, P = 0.02) and HSG5 (adjusted HR = 2.91, 1.77–4.79, p < 0.001) (Fig. 4). Sensitivity analyses with alternative imputation methods yielded consistent results (Fig. S4). CumFWR and trajectories similarly predicted SAP, persistent MOF, and IPN (Fig. S10).
Fig. 4.
Cox Regression Forest Plot. Multiple models were constructed using a stratified adjustment strategy: Model 1: Unadjusted. Model 2: Model 1 + Sex, Age, Smoking History, Alcohol History. Model 3: Model 2 + Etiology (Biliary, Alcoholic, Hypertriglyceridemic, Other/Idiopathic), Body Temperature, Heart Rate. Model 4: Model 3 + Baseline Laboratory Indicators (e.g., Admission HCT, BUN, TG)
Three-dimensional modeling revealed synergistic mortality risk amplification when high CumFWR coincided with elevated APACHE II scores (> 20) and CumFWR > 100 mL/kg (Fig. S11). Admission hematocrit (HCT) exhibited a nonlinear association with mortality (p < 0.001), where both low and high HCT increased risk (Fig. 5A). The HCT-CumFWR interaction surface showed peak mortality risk at combined high CumFWR with low or high HCT (Fig. 5B). This relationship persisted across trajectory groups (Fig. S12), with minimal impact in LSG1/LMSG2. In MSG3, MHFG4, and HSG5, high CumFWR with HCT extremes conferred higher risk.
Fig. 5.
Interaction Effect Between HCT and Fluid Therapy. A Dose–effect relationship between Admission Hematocrit (HCT) and Mortality Risk (RCS model). B Joint effect of HCT and CumFWR on Mortality Risk (3D fitted surface)
Propensity score matching (PSM) analysis
To address confounding, we performed 1:1 propensity score matching between high-volume trajectories (HVT: MHFG4 + HSG5) and low-moderate volume trajectories (LMVT: LSG1 + LMSG2 + MSG3), yielding 483 matched pairs. Post-matching analysis demonstrated excellent covariate balance (SMD < 0.1 for all variables; Tables S6, S8; Fig. S5), with comparable CumFWR distributions across different SIRS and severity (Fig. S6). The HVT group maintained significantly higher complication rates and mortality (Figs. 6 and 7, Table S9), consistent across alternative weighting methods (Table S7). Subgroup analyses revealed particularly elevated HVT-associated mortality in biliary pancreatitis, males, BMI ≥ 24 kg/m2, SIRS ≥ 2, and early (≤ 28-day) mortality (Fig. 8).
Fig. 6.
Kaplan–Meier Survival Curves. Within 28 days (A), Within 90 days (B), During Hospitalization (C)
Fig. 7.
Forest Plot After Propensity Score Matching. MOF, multiple organ failure; IPN, infected pancreatic necrosis
Fig. 8.
Subgroup Analysis Forest Plot After Propensity Score Matching
Sensitivity analysis
Fourteen sensitivity analyses confirmed HVT mortality risk persistence, including: Temporal cohort restriction (excluding 2005–2013 data); Comorbidity exclusion (hypertension/COPD/diabetes); Early admission subset (≤ 24 h post-onset); First-episode AP restriction; Severe disease subset (APACHE II > 8); Enhanced covariate adjustment (MAP, PaO2/FiO2, albumin/LMWH/insulin) (Fig. S7). Unmeasured confounding analysis (E-value = 4.05) indicated that only strong confounders (HR ≥ 4.05 for both exposure/outcome) could nullify the association (Fig. S8, Table S10).
Discussion
This study reveals the complex association between the spatiotemporal heterogeneity of early fluid therapy and prognosis in AP by integrating cumulative fluid load and dynamic trajectory models. Based on cohort data of 3142 AP patients, we not only confirmed the relationship between cumulative FWR and mortality risk, but more importantly, employed LCGM to identify five distinct fluid therapy trajectories within 72 h, demonstrating significantly higher mortality risk in the MHFG4 and HSG5 groups compared to the other three trajectory. This highlights the clinical need to move beyond simplistic volume control management and prioritize close monitoring of temporal characteristics during fluid therapy.
Fluid resuscitation is a cornerstone intervention in AP, primarily aimed at maintaining effective tissue perfusion, improving microcirculation, and mitigating pancreatic/systemic inflammation [26–28]. However, substantial variation persists in clinical strategies regarding fluid composition, infusion rates, and goal-directed protocols [29–31]. Traditional early aggressive fluid resuscitation approaches risk fluid overload, potentially increasing complications like pulmonary edema and ACS [32, 33]. This study revealed significant heterogeneity in FWR among AP patients. The high-sustained group (Group 5) exhibited the worst prognosis, with significantly increased rates of SAP, IPN, MPOF, and in-hospital mortality. These adverse outcomes likely stem from fluid overload cascades, consistent with the WATERFALL trial's finding that excessive resuscitation exacerbates organ injury while controlled volumes improve outcomes [7]. Mechanistically, excessive crystalloid infusion dilutes plasma proteins, impairing anti-inflammatory mediators (e.g., albumin) and triggering proinflammatory cytokine surges (IL-6, TNF-α), thereby exacerbating SIRS [34]. This process is similar to the results of studies on fluid management in septic patients [35]. Fluid retention also elevates intra-abdominal pressure, potentially inducing ACS that compromises intestinal microcirculation through mechanical compression, worsening ischemia–reperfusion injury and bacterial translocation [36, 37].
A meta-analysis demonstrated that critically ill AP patients receiving high-volume intravenous fluid resuscitation faced a 2.45-fold increased mortality risk compared to non-aggressive regimens [38]. This study quantified the synergistic risk through 3D modeling, when APACHE II > 20 and CumFWR > 100 mL/kg, mortality increased exponentially. Nonlinear RCS curves confirmed elevated mortality at both low HCT (hemodilution) and high HCT (hemoconcentration). The 3D interaction model showed high CumFWR amplified this risk, particularly at HCT extremes. Trajectory stratification identified maximal risk amplification in the HSG5 group (p < 0.001), indicating fundamental differences in hemorheological tolerance across therapy patterns. These findings establish a theoretical basis for individualized fluid management [39]. In high-load trajectories (HSG5/MHFG4), HCT should be dynamically maintained within 35–44% to prevent oxygen delivery failure from hemodilution or microcirculatory collapse from hemoconcentration [3].
This study provides important implications for clinical practice. Fluid management should adopt a dynamic approach, with daily assessment of FWR trends during the first 72 h of hospitalization [28, 40]. For those with fluctuating or sustained high loads, intervention measures should be taken as early as possible, such as bedside ultrasound-guided fluid challenge or use of vasoactive drugs to reduce fluid demand. It is recommended to take CumFWR 100 mL/kg as the early warning threshold for severe AP, especially when APACHE II > 20, a restrictive fluid strategy (such as goal-directed hourly balance monitoring) should be initiated [29, 41]. HCT should be used as a key dynamic marker of fluid responsiveness. Maintaining HCT in the 35–44% range in high-risk trajectory groups may improve prognosis, in cases of high HCT, diuresis should be strengthened instead of blind fluid replacement. In cases of low HCT, further crystalloid expansion should be used with extreme caution. Rather than automatic fluid boluses, individualized assessment of volume status and end-organ perfusion (e.g., via ultrasound, lactate, or central venous oxygen saturation) is recommended [3, 5, 41–43]. Transfusion may be considered only in the context of symptomatic anemia with evidence of impaired oxygen delivery, in line with current guidelines. However, this approach requires validation in prospective studies.
Although this study introduces novel analytical perspectives on fluid therapy trajectories, several limitations should also be recognized. Firstly, the retrospective design may introduce confounding bias. It is worth noting that the prognostic discrimination of trajectory grouping remains robust after PSM and multiple weighting analyses, and its robustness is further supported by unmeasured confounding analysis through 14 sensitivity analyses. Secondly, the LCGM model does not include changes in the ratio of colloids/crystalloids, and future trajectory analysis needs to be deepened in combination with fluid types; thirdly, this study only observed outcome indicators during hospitalization, lacking follow-up studies on patients' long-term prognosis. In addition, prospective, multi-center, large-sample studies are needed in the future to further verify the clinical utility of CumFWR and dynamic trajectory models in fluid management of AP patients.
In conclusion, this study demonstrates the clinical value of dynamic monitoring of early fluid therapy in AP through CumFWR and trajectory modeling. Critically, the CumFWR-HCT three-dimensional risk assessment establishes a stratified management framework for AP patients. This facilitates the evolution of clinical practice from volume control to physiology-guided dynamic optimization—prioritizing treatment timing and individualized physiological response.
Supplementary Information
Acknowledgements
The authors thank the Jiangxi Clinical Research Center for Digestive Disease (no. 20201ZDG020007) for their support as well as all the participants who were involved in this study. We thank Free Statistics team for providing technical assistance and valuable tools for data analysis and visualization.
Abbreviations
- AP
Acute pancreatitis
- SAP
Severe acute pancreatitis
- LMWH
Low molecular weight heparin
- MODS
Multiple organ dysfunction syndrome
- SIRS
Systemic inflammatory response syndrome
- PSM
Propensity score matching
- IPTW
Inverse probability of treatment weighting
- OW
Overlap weighting
- TG
Triglyceride
- BMI
Body mass index
- APACHE II
Acute Physiology and Chronic Health Evaluation II
- HR
Hazard ratio
- CI
Confidence interval
- SMD
Standardized mean difference
- DAGs
Directed acyclic graphs
- FWR
Fluid therapy-to-body weight ratio
- CumFWR
Cumulative fluid therapy-to-body weight ratio
- LCGM
Latent Class Growth Modeling
- HCT
Hematocrit
- BUN
Blood urea nitrogen
- POF
Persistent organ failure
- IPN
Infected pancreatic necrosis
- ICU
Intensive care unit
- RRT
Renal replacement therapy
- CECT
Contrast-enhanced CT
- RCS
Restricted cubic splines
- AUC
Area under the curve
- LSG1
Low Stable Group 1
- LMSG2
Low-Moderate Sustained Group 2
- MSG3
Moderate Stable Group 3
- MHFG4
Moderate-High Fluctuating Group 4
- HSG5
High Sustained Group 5
- HVT
High Volume Trajectory Group
- LMVT
Low-Moderate Volume Trajectory Group
- STROBE
Strengthening the Reporting of Observational Studies in Epidemiology
- PMOF
Persistent multiple organ failure
- IPTW
Inverse probability of treatment weighting
- SMRW)
Standardized mortality ratio weighting
- PA
Pairwise algorithmic weighting
- OW
Overlap weighting
- MAP
Mean arterial pressure
- PaO2/FiO2
Oxygenation index
- ACS
Abdominal compartment syndrome
Author contributions
Jianhua Wan: Data curation, Writing—original draft, Validation, Formal analysis, Conceptualization, Funding acquisition. Shixuan Xiong: Writing—review & editing, Methodology, Investigation, Formal analysis. Yaoyu Zou: Writing—review & editing, Methodology, Investigation, Formal analysis. Maobin Kuang: Writing—review & editing, Methodology, Formal analysis. Huajing Ke: Writing—review & editing, Methodology, Formal analysis. Wenhua He: Writing—review & editing, Conceptualization. Yin Zhu: Writing—review & editing, Supervision. Nonghua Lu: Writing—review & editing, Supervision. Liang Xia: Writing—review & editing, Conceptualization, Funding acquisition, Supervision.
Funding
The study design, data collection, analysis and interpretation of data, writing manuscript were funded by the National Natural Science Foundation of China (No: 82560136 and 82560137). This study was supported by the Project for Academic and Technical Leaders of Major Disciplines in Jiangxi Province (20243BCE51144), Jiangxi Provincial Natural Science Foundation (20242BAB25438) and The First Affiliated Hospital of Nanchang University Clinical Research and Cultivation Project (YFYKCTDPY202202).
Data availability
All data generated or analyzed during this study are included in this published article.
Declarations
Ethics approval and consent to participate
The study was approved by the ethics committee of The First Affiliated Hospital of Nanchang University (No. 2011001).
Consent for publication
All authors have approved the manuscript and consented to its submission and publication.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Jianhua Wan, Shixuan Xiong, and Yaoyu Zou equal contribution of first authors.
References
- 1.Petrov MS, Yadav D. Global epidemiology and holistic prevention of pancreatitis. Nat Rev Gastroenterol Hepatol. 2019;16:175–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Banks PA, Bollen TL, Dervenis C, et al. Classification of acute pancreatitis–2012: revision of the Atlanta classification and definitions by international consensus. Gut. 2013;62:102–11. [DOI] [PubMed] [Google Scholar]
- 3.info@internationalpancreatology.org IAEIJWGEa, Group IAEIJW. International Association of Pancreatology Revised Guidelines on Acute Pancreatitis 2025. Pancreatology 2025.
- 4.Trikudanathan G, Yazici C, Evans Phillips A, et al. Diagnosis and management of acute pancreatitis. Gastroenterology. 2024;167:673–88. [DOI] [PubMed] [Google Scholar]
- 5.Tenner S, Vege SS, Sheth SG, et al. American College of Gastroenterology guidelines: management of acute pancreatitis. Am J Gastroenterol. 2024;119:419–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Yang AL, Vege SS. Fluid resuscitation in acute pancreatitis. Curr Opin Gastroenterol. 2023;39:411–5. [DOI] [PubMed] [Google Scholar]
- 7.de-Madaria E, Buxbaum JL, Maisonneuve P, et al. Aggressive or moderate fluid resuscitation in acute pancreatitis. N Engl J Med. 2022;387:989–1000. [DOI] [PubMed] [Google Scholar]
- 8.Messmer AS, Zingg C, Muller M, Gerber JL, Schefold JC, Pfortmueller CA. Fluid overload and mortality in adult critical care patients-a systematic review and meta-analysis of observational studies. Crit Care Med. 2020;48(12):1862–70. 10.1097/CCM.0000000000004617. [DOI] [PubMed] [Google Scholar]
- 9.De Waele JJ, Ejike JC, Leppaniemi A, et al. Intra-abdominal hypertension and abdominal compartment syndrome in pancreatitis, paediatrics, and trauma. Anaesthesiol Intensive Ther. 2015;47:219–27. [DOI] [PubMed] [Google Scholar]
- 10.Trikudanathan G, Navaneethan U, Vege SS. Current controversies in fluid resuscitation in acute pancreatitis: a systematic review. Pancreas. 2012;41:827–34. [DOI] [PubMed] [Google Scholar]
- 11.Beij A, Verdonk RC, van Santvoort HC, et al. Acute pancreatitis: an update of evidence-based management and recent trends in treatment strategies. United Eur Gastroenterol J. 2025. 10.1002/ueg2.12743. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Machicado JD, Papachristou GI. Intravenous fluid resuscitation in the management of acute pancreatitis. Curr Opin Gastroenterol. 2020;36:409–16. [DOI] [PubMed] [Google Scholar]
- 13.Crosignani A, Spina S, Marrazzo F, et al. Intravenous fluid therapy in patients with severe acute pancreatitis admitted to the intensive care unit: a narrative review. Ann Intensive Care. 2022;12:98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Bhatraju PK, Wurfel MM, Himmelfarb J. Trajectory of kidney function: the canary in sepsis. Am J Respir Crit Care Med. 2020;202:1211–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Nagin DS, Odgers CL. Group-based trajectory modeling in clinical research. Annu Rev Clin Psychol. 2010;6:109–38. [DOI] [PubMed] [Google Scholar]
- 16.Takkavatakarn K, Oh W, Chan L, et al. Machine learning derived serum creatinine trajectories in acute kidney injury in critically ill patients with sepsis. Crit Care. 2024;28:156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.White KC, Serpa-Neto A, Hurford R, et al. Sepsis-associated acute kidney injury in the intensive care unit: incidence, patient characteristics, timing, trajectory, treatment, and associated outcomes. A multicenter, observational study. Intensive Care Med. 2023;49:1079–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Holm S. Declaration of helsinki. International encyclopedia of ethics 2013:1–4.
- 19.Cuschieri S. The STROBE guidelines. Saudi J Anaesth. 2019;13:S31–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Izawa J, Kitamura T, Iwami T, et al. Early-phase cumulative hypotension duration and severe-stage progression in oliguric acute kidney injury with and without sepsis: an observational study. Crit Care. 2016;20:1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Domanski MJ, Tian X, Wu CO, Reis JP, Dey AK, Gu Y, et al. Time course of LDL cholesterol exposure and cardiovascular disease event risk. J Am Coll Cardiol. 2020;76(13):1507–16. 10.1016/j.jacc.2020.07.059. [DOI] [PubMed] [Google Scholar]
- 22.Andruff H, Carraro N, Thompson A, et al. Latent class growth modelling: a tutorial. Tutorials Quant Methods Psychol. 2009;5:11–24. [Google Scholar]
- 23.Feeney T, Hartwig FP, Davies NM. How to use directed acyclic graphs: guide for clinical researchers. BMJ. 2025;388:e078226. [DOI] [PubMed] [Google Scholar]
- 24.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167:268–74. [DOI] [PubMed] [Google Scholar]
- 25.Luo N, Zhang X, Huang J, Chen H, Tang H. Prevalence of steatotic liver disease and associated fibrosis in the United States: results from NHANES 2017-March 2020. J Hepatol. 2024;80(2):e70–1. 10.1016/j.jhep.2023.08.019. [DOI] [PubMed] [Google Scholar]
- 26.Hines OJ, Pandol SJ. Management of severe acute pancreatitis. BMJ. 2019;367:l6227. [DOI] [PubMed] [Google Scholar]
- 27.He K, Gao L, Yang Z, Zhang Y, Hua T, Hu W, et al. Aggressive versus controlled fluid resuscitation in acute pancreatitis: a systematic review and meta-analysis of randomized controlled trials. Chin Med J (Engl). 2023;136(10):1166–73. 10.1097/CM9.0000000000002684. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.James TW, Crockett SD. Management of acute pancreatitis in the first 72 hours. Curr Opin Gastroenterol. 2018;34:330–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Froghi F, Soggiu F, Ricciardi F, Vindrola-Padros C, Floros L, Martin D, et al. Ward based goal directed fluid therapy (GDFT) in acute pancreatitis (GAP) trial: a feasibility randomised controlled trial. Int J Surg. 2022;104:106737. 10.1016/j.ijsu.2022.106737. [DOI] [PubMed] [Google Scholar]
- 30.Dawson A, Karunakaran M, Sharma ZD, Ullah S, Barreto SG. Fluid resuscitation in the early management of acute pancreatitis - evidence from a systematic review and meta-analysis. HPB (Oxford). 2023;25(12):1451–65. 10.1016/j.hpb.2023.08.013. [DOI] [PubMed] [Google Scholar]
- 31.Liu B, Ding X, Yang J. Effect of early goal directed therapy in the treatment of severe sepsis and/or septic shock. Curr Med Res Opin. 2016;32:1773–82. [DOI] [PubMed] [Google Scholar]
- 32.van Brunschot S, Schut AJ, Bouwense SA, et al. Abdominal compartment syndrome in acute pancreatitis: a systematic review. Pancreas. 2014;43:665–74. [DOI] [PubMed] [Google Scholar]
- 33.De Waele JJ, Leppaniemi AK. Intra-abdominal hypertension in acute pancreatitis. World J Surg. 2009;33:1128–33. [DOI] [PubMed] [Google Scholar]
- 34.Malbrain ML, Marik PE, Witters I, 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] [PubMed] [Google Scholar]
- 35.Silversides JA, Major E, Ferguson AJ, et al. Conservative fluid management or deresuscitation for patients with sepsis or acute respiratory distress syndrome following the resuscitation phase of critical illness: a systematic review and meta-analysis. Intensive Care Med. 2017;43:155–70. [DOI] [PubMed] [Google Scholar]
- 36.Elvevoll B, Husby P, Ovrebo K, et al. Acute elevation of intra-abdominal pressure contributes to extravascular shift of fluid and proteins in an experimental porcine model. BMC Res Notes. 2014;7:738. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Jacobs R, Wise RD, Myatchin I, et al. Fluid management, intra-abdominal hypertension and the abdominal compartment syndrome: a narrative review. Life (Basel). 2022. 10.3390/life12091390. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Li XW, Wang CH, Dai JW, et al. Comparison of clinical outcomes between aggressive and non-aggressive intravenous hydration for acute pancreatitis: a systematic review and meta-analysis. Crit Care. 2023;27:122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Wang Z, Wang W, Xu J, et al. Development and validation of dynamic clinical subphenotypes in acute pancreatitis patients using vital sign trajectories in intensive care units: a multinational cohort study. Signal Transduct Target Ther. 2025;10:180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Kong N, Chang P, Ira S, et al. Machine learning-guided fluid resuscitation for acute pancreatitis improves outcomes. Clin Transl Gastroenterol 2025. [DOI] [PMC free article] [PubMed]
- 41.Mekontso Dessap A, AlShamsi F, Belletti A, et al. European Society of Intensive Care Medicine (ESICM) 2025 clinical practice guideline on fluid therapy in adult critically ill patients: part 2-the volume of resuscitation fluids. Intensive Care Med. 2025;51:461–77. [DOI] [PubMed] [Google Scholar]
- 42.Yaowmaneerat T, Sirinawasatien A. Update on the strategy for intravenous fluid treatment in acute pancreatitis. World J Gastrointest Pharmacol Ther. 2023;14:22–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Li L, Jin T, Wen S, et al. Early rapid fluid therapy is associated with increased rate of noninvasive positive-pressure ventilation in hemoconcentrated patients with severe acute pancreatitis. Dig Dis Sci. 2020;65:2700–11. [DOI] [PMC free article] [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
All data generated or analyzed during this study are included in this published article.