Abstract
This retrospective study evaluated the impact of intraosseous infusion (IO) versus traditional intravenous infusion (IV) on 30-day mortality and clinical outcomes in 518 patients with acute gastrointestinal bleeding (AGIB) secondary to gastrointestinal tumors from January 2022 to July 2024. Patients were divided into IO (n=217) and IV (n=301) groups based on initial resuscitation strategy. Compared to IV group, the IO group demonstrated higher first-attempt catheterization success rate, shorter vascular access time, and faster blood pressure recovery (all P<0.001), alongside higher 6-hour lactate (LA) clearance (34% vs. 22%, P<0.001) and lower 30-day mortality (11.98% vs. 18.6%, P=0.016). Multivariate analysis identified IO infusion as protective factor for lactate metabolism (HR=0.289, 95% CI: 0.092-0.864), while advanced age (HR=1.125), diabetes (HR=3.23), and low LA clearance (HR=0.016) were independent risk factor for mortality. Causal mediation analysis revealed that 6-hour LA clearance mediated 68% of the IO-associated mortality reduction (P<0.001), whereas diabetes history was not a significant mediator (P=0.156). Complication rates were comparable between groups (P>0.05). These findings indicate that IO infusion improves survival in AGIB due to gastrointestinal tumors by rapidly restoring hemodynamics and enhancing lactate metabolism. The mortality benefit is primarily driven by accelerated LA clearance rather than comorbidities like diabetes. Given its safety profile comparable to IV, IO infusion should be prioritized in critical care settings.
Keywords: Gastrointestinal tumor, acute massive hemorrhage, intraosseous infusion, intravenous infusion, 30-day mortality, lactate clearance, prognostic factors
Introduction
Gastrointestinal (GI) tumors rank among the most prevalent malignancies globally, with rising incidence rates, particularly in regions where gastric and esophageal cancers are common [1]. Global cancer burden data indicate that gastric and esophageal cancers remain major contributors to cancer-related mortality [2]. Despite advances in early detection and treatment, patients with advanced GI tumors often experience severe complications, such as acute gastrointestinal bleeding (AGIB), which significantly threatens survival and quality of life [3].
AGIB is a common and potentially fatal complication in patients with GI malignancies. In advanced stages of gastric, esophageal, or other GI cancers, tumor invasion into blood vessels or ulcer formation can lead to massive hemorrhage [4], resulting in hypovolemia, hypotension, electrolyte disturbances, shock, and multi-organ failure [5]. Prompt fluid resuscitation and blood transfusion are therefore vital, as delays may lead to a 30-day mortality rate ranging from 30% to 40% [6]. Thus, early and effective management is essential for improving clinical outcomes.
Intravenous (IV) infusion is the standard method for fluid resuscitation. However, it presents challenges in AGIB patients [7]. Peripheral vasoconstriction, hypoperfusion, and hypotension often impair IV catheterization success, and infusion rates may be inadequate for rapid volume replacement [8]. Intraosseous (IO) infusion has emerged as a promising alternative, providing rapid and reliable vascular access via the bone marrow cavity and enabling efficient fluid delivery in emergency and critical care settings [9].
IV access is particularly difficult in patients with hemorrhagic shock or profound hypotension due to collapsed peripheral veins [10]. Delays in achieving vascular access and initiating resuscitation can worsen patient outcomes and elevate mortality risk. In contrast, IO infusion overcomes peripheral vascular limitations by enabling rapid fluid administration through bone marrow access [11]. It has been increasingly adopted in emergency medicine and critical care, particularly for scenarios such as cardiac arrest and acute hemorrhage [12]. However, limited evidence exists on its effectiveness in GI tumor-related AGIB, and its impact on 30-day mortality remains underexplored.
This study retrospectively compared the clinical efficacy of IO versus IV infusion in patients with GI tumors complicated by AGIB, with a primary focus on 30-day mortality. By assessing outcomes such as resuscitation efficiency, lactate clearance, and survival, we seek to determine whether IO infusion offers a more superior alternative in this high-risk population.
Methods and materials
Sample size calculation
Based on the study by Gong et al. [13], which reported a 30-day mortality rate of 17.8% in cancer-related non-variceal upper gastrointestinal bleeding treated with transarterial embolization, the sample size was calculated using the formula: N = Z2 × [P × (1 - P)]/E2, where E=0.05, Z=1.96, and P=0.178. The estimated minimum sample size was 225 patients, with the final sample size adjusted based on clinical availability.
Patient selection
This retrospective study included 518 patients with gastrointestinal (GI) tumors complicated by active gastrointestinal bleeding (AGIB), admitted between January 2022 and July 2024. The study was approved by the ethics committee of The People’s Hospital of Rugao (Figure 1).
Figure 1.
Sample collection flow chart.
Inclusion and exclusion criteria
Inclusion criteria: (1) Confirmed diagnosis of a GI tumor (e.g., esophageal, gastric, colorectal, or other GI tract tumors) with concurrent AGIB; (2) Age ≥18 years; (3) Definitive AGIB diagnosis confirmed by clinical symptoms, endoscopy, and/or imaging; (4) Treatment initiated within 72 hours of bleeding onset; (5) Complete clinical records.
Exclusion criteria: (1) History of severe cardiac disease (e.g., advanced heart failure or myocardial infarction); (2) Inability to assess conditions (e.g., unconsciousness or poor cooperation); (3) Concurrent malignancies affecting outcome interpretation; (4) Unidentified bleeding source despite endoscopy or imaging; (5) Pregnant or lactating women due to potential treatment risks.
Definition of AGIB
AGIB was defined as an acute blood loss of ≥500 mL with significant physiological impact. Common clinical presentations included hematemesis (vomiting of bright red or coffee-ground-like blood), melena (black, tarry stools), or hematochezia (passage of fresh blood per rectum). Melena typically indicates upper GI bleeding, while hematochezia suggests lower GI sources [14].
Fluid resuscitation protocols
IV group: Central venous access was established via the external jugular or subclavian vein (1 cm below the midclavicular point). Following local anesthesia with 2% lidocaine (2 mL), a puncture needle was inserted, and successful venous entry was confirmed by blood return. A guidewire was introduced, the skin was dilated, and a catheter was placed and connected to the infusion device.
IO group: Patients were positioned supine, and the puncture site was selected 1-3 cm below the tibial tuberosity on the medial flat surface. After standard disinfection, an IO needle was inserted into the bone marrow at a 90° angle using a power driver. Entry was confirmed by loss of resistance and bone marrow aspiration. A pre-flushed connector was attached, followed by a 10 mL saline bolus and connection to the infusion device. After 24 hours, IO access was transitioned to conventional IV infusion per standard clinical protocol [11].
Data collection
Baseline variables included demographics (age, gender, BMI), tumor type, TNM stage, estimated 24-hour blood loss, time from bleeding onset to admission, hemorrhagic shock status, and comorbidities (diabetes, hypertension, prior radiotherapy/chemotherapy). Treatment details encompassed fluid resuscitation method (IV vs. IO), hemostatic intervention (endoscopy vs. embolization), first-attempt catheterization success rate, time to vascular access, time to blood pressure recovery, infusion rate, and urine output. Laboratory data involved pre- and 6-hour post-infusion levels of lactate (LA) and albumin (ALB); lactate clearance rate = [(pre-infusion LA - 6-hour post-infusion LA)/pre-infusion LA] × 100%. Outcomes included 30-day all-cause mortality, complication rates (local swelling, catheter dislodgement, infection, fluid extravasation), and resuscitation success rate.
All data were extracted from the hospital information system (HIS), including admission notes, progress records, laboratory results, nursing records, and discharge summaries.
Laboratory testing
Lactate (LA) and albumin (ALB) levels were measured using the Beckman Coulter AU5800 automatic biochemical analyzer (Beckman Coulter, USA). Peripheral venous blood (5 mL) was collected before and 6 hours after infusion into heparinized tubes and then centrifuged at 3000 rpm for 10 minutes (radius 15 cm) to isolate plasma. LA was assayed using the lactate oxidase colorimetric method (kit OSR6120, λ=540 nm), and ALB was measured using the bromocresol green method (kit OSR6102, λ=628 nm). Lactate clearance was calculated as described above and reported in mmol/L for LA and g/L for ALB.
Outcome definitions
Hemorrhagic death: Caused by AGIB-related events (e.g., hypovolemic shock, multi-organ failure, refractory bleeding).
Non-hemorrhagic death: Attributed to tumor progression, septic shock, or cardiopulmonary failure.
Cause of death was determined based on electronic health record (EHR) documentation and death certificates.
Outcome measures
Primary outcomes
Risk factors for low lactate clearance (<10%); independent predictors of 30-day mortality.
Secondary outcomes
First-attempt catheterization success rate and overall resuscitation success; time to establish infusion access, blood pressure recovery, infusion rate, and urine output; changes in LA and ALB levels before and 6 hours after infusion; 6-hour LA clearance; complications (e.g., swelling, catheter dislodgement, infection, extravasation).
Statistical analysis
Data were analyzed using SPSS version 27.0 and R version 4.3.3. Categorical variables were reported as frequencies and percentages and compared using the chi-square test or Correction should test, as appropriate. Continuous variables were assessed for normality with the Kolmogorov-Smirnov test. Normally distributed data were expressed as mean ± standard deviation and analyzed using independent-samples or paired-samples t-tests. Non-normally distributed data were reported as medians with interquartile ranges and compared using the Mann-Whitney U test or Wilcoxon signed-rank test.
Logistic regression was used to identify independent risk factors for low lactate clearance (<10%). Cox proportional hazards regression models were applied to analyze factors associated with 30-day mortality, with hazard ratios (HRs) and 95% confidence intervals (CIs) calculated. Receiver operating characteristic (ROC) curves and area under the curve (AUC) values were generated using the pROC package, with differences compared via DeLong’s test. Kaplan-Meier survival curves and forest plots were constructed using the survival package, with group comparisons assessed by the log-rank test.
Competing risk analysis was conducted using the Fine-Gray model with the cmprsk package to account for non-hemorrhagic death as a competing event. Mediation analysis was performed using the mediation package in R to explore mechanisms between independent variables and outcomes through potential mediators, with standardized path coefficients assessing mediation effect magnitude.
All models were tested for appropriate assumptions, and confidence intervals were computed. All hypothesis tests were two-sided, with a P value <0.05 considered statistically significant.
Results
Comparison of baseline characteristics between the IV and IO groups
Baseline characteristics were well-balanced between patients with GI tumor-related massive hemorrhage receiving IO and IV treatments. No significant differences were observed between the IO and IV groups in age (P=0.411), gender (P=0.560), body mass index (P=0.238), cancer type (P=0.753), TNM stage (P=0.580), 24-hour blood loss (P=0.374), transfusion therapy within 24 hours (P=0.261), time from bleeding onset to admission (P=0.358), occurrence of hemorrhagic shock (P=0.560), history of chemotherapy/radiotherapy (P=0.456), diabetes (P=0.153), or hypertension (P=0.497), indicating strong comparability between groups (Table 1).
Table 1.
Comparison of baseline characteristics between IO group and IV group
| Variable | Total | IV Group (n=301) | IO Group (n=217) | Statistic | P-Value |
|---|---|---|---|---|---|
| Age (years) | 65.12±7.84 | 65.36±7.73 | 64.79±7.99 | 0.822 | 0.411 |
| Gender | |||||
| Male | 384 (74.13%) | 226 (75.08%) | 158 (72.81%) | 0.339 | 0.560 |
| Female | 134 (25.87%) | 75 (24.92%) | 59 (27.19%) | ||
| BMI (kg/m2) | 24.00±2.99 | 24.13±3.00 | 23.81±2.97 | 1.180 | 0.238 |
| Cancer Type | |||||
| Gastric Cancer | 305 (58.88%) | 181 (60.13%) | 124 (57.14%) | 0.567 | 0.753 |
| Esophageal Cancer | 185 (35.71%) | 105 (34.88%) | 80 (36.87%) | ||
| Other | 28 (5.41%) | 15 (4.98%) | 13 (5.99%) | ||
| TNM Stage | |||||
| Stage II | 56 (10.81%) | 30 (9.97%) | 26 (11.98%) | 1.091 | 0.580 |
| Stage III | 134 (25.87%) | 75 (24.92%) | 59 (27.19%) | ||
| Stage IV | 328 (63.32%) | 196 (65.12%) | 132 (60.83%) | ||
| 24 h Blood Loss | |||||
| ≥1000 mL | 105 (20.27%) | 57 (18.94%) | 48 (22.12%) | 0.790 | 0.374 |
| <1000 mL | 413 (79.73%) | 244 (81.06%) | 169 (77.88%) | ||
| 24 h Blood Transfusion | |||||
| Yes | 281 (54.25%) | 157 (52.16%) | 124 (57.14%) | 1.262 | 0.261 |
| No | 237 (45.75%) | 144 (47.84%) | 93 (42.86%) | ||
| Time from Bleeding to Admission | |||||
| ≥12 h | 315 (60.81%) | 178 (59.14%) | 137 (63.13%) | 0.846 | 0.358 |
| <12 h | 203 (39.19%) | 123 (40.86%) | 80 (36.87%) | ||
| Hemorrhagic Shock | |||||
| Yes | 134 (25.87%) | 75 (24.92%) | 59 (27.19%) | 0.339 | 0.560 |
| No | 384 (74.13%) | 226 (75.08%) | 158 (72.81%) | ||
| Received Radiation/Chemotherapy | |||||
| Yes | 346 (66.80%) | 205 (68.11%) | 141 (64.98%) | 0.557 | 0.456 |
| No | 172 (33.20%) | 96 (31.89%) | 76 (35.02%) | ||
| Diabetes History | |||||
| Yes | 104 (20.08%) | 54 (17.94%) | 50 (23.04%) | 2.045 | 0.153 |
| No | 414 (79.92%) | 247 (82.06%) | 167 (76.96%) | ||
| Hypertension History | |||||
| Yes | 187 (36.10%) | 105 (34.88%) | 82 (37.79%) | 0.461 | 0.497 |
| No | 331 (63.90%) | 196 (65.12%) | 135 (62.21%) | ||
| Hemostasis Method | |||||
| Interventional Embolization | 333 (64.29%) | 137 (63.13%) | 196 (65.12%) | 0.216 | 0.642 |
| Endoscopic Hemostasis | 185 (35.71%) | 80 (36.87%) | 105 (34.88%) |
Note: BMI: Body Mass Index.
Comparison of first-attempt cannulation and resuscitation success rates between the IV and IO groups
The IO group exhibited a significantly higher first-attempt cannulation success rate compared to the IV group (P<0.001, Figure 2A), demonstrating greater efficiency in emergency vascular access. However, no significant difference was found in resuscitation success rates between the two groups (P=0.194, Figure 2B).
Figure 2.
Comparison of first-attempt cannulation success rate and resuscitation success rate between IO group and IV group. A. First-attempt cannulation success rate; B. Resuscitation success rate. Note: IO: Intraosseous infusion, IV: Intravenous infusion.
Comparison of fluid-related indicators between the IV and IO groups
The IO group required significantly less time to establish an infusion channel (P<0.001, Figure 3A) and achieved faster blood pressure recovery after fluid resuscitation (P<0.001, Figure 3B) compared to the IV group. No significant differences were observed in infusion rate (P=0.823, Figure 3C) or urine output (P=0.466, Figure 3D) between the groups.
Figure 3.
Comparison of fluid-related indicators between IO group and IV group. A. Time required for establishing an infusion pathway; B. Time required for blood pressure recovery after fluid resuscitation; C. Infusion rate; D. Urine output. Note: IO: Intraosseous infusion, IV: Intravenous infusion; “ns” indicates no significant difference, “***” indicates P<0.001.
Comparison of lactate (LA) and albumin (ALB) levels between the two groups before and after infusion
Before fluid infusion, LA and ALB levels showed no significant differences between the IO and IV groups (P>0.05). Post-infusion, both groups exhibited significant reductions in LA levels (P<0.001) and increases in ALB levels (P<0.001). The IO group demonstrated significantly greater reductions in LA and increases in ALB compared to the IV group (P<0.001, Figure 4).
Figure 4.
Comparison of changes in LA and ALB levels between IO and IV groups before and after infusion. A. Change in LA levels before and 6 hours after infusion; B. Change in ALB levels before and 6 hours after infusion. Note: LA: Lactate, ALB: Albumin, IO: Intraosseous infusion, IV: Intravenous infusion; “ns” indicates no significant difference, “***” indicates P<0.001.
Comparison of lactate clearance rate between the two groups
The IO group achieved a significantly higher 6-hour lactate clearance rate than the IV group (P<0.001, Figure 5).
Figure 5.
Comparison of 6-hour LA clearance rate between IO and IV groups. Note: LA: Lactate, IO: Intraosseous infusion, IV: Intravenous infusion; “***” indicates P<0.001.
Comparison of incidence of complications between the two groups
No significant differences in complication rates were observed between the IV and IO groups. Local swelling occurred in 24 cases in the IV group and 15 in the IO group (P=0.651). Catheter dislodgement was reported in 12 cases in the IV group and 9 in the IO group (P=0.927). Infections occurred in 5 cases in the IV group and 4 in the IO group (P=0.854). Fluid extravasation was noted in 12 cases in the IV group and 7 in the IO group (P=0.649) (Table 2).
Table 2.
Comparison of complication incidence between IV group and IO group
| Group | Local Swelling | Catheter Dislodgement | Infection | Fluid Extravasation |
|---|---|---|---|---|
| IV Group (n=301) | 24 | 12 | 5 | 12 |
| IO Group (n=217) | 15 | 9 | 4 | 7 |
| Chi-square Value | 0.204 | 0.008 | 0.034 | 0.207 |
| P Value | 0.651 | 0.927 | 0.854 | 0.649 |
Note: IV: Intravenous, IO: Intraosseous.
Comparison of baseline characteristics between patients stratified by lactate clearance rate
Older age (P<0.001), 24-hour blood loss ≥1000 mL (P=0.004), time from bleeding onset to admission ≥12 hours (P=0.021), and hemorrhagic shock (P=0.030) were significantly associated with a lactate clearance rate <10%. The proportion of patients with a lactate clearance rate ≥10% was significantly higher in the IV group than in the IO group (P=0.003). Additionally, patients requiring ≥12 hours for blood pressure recovery after fluid resuscitation had a significantly higher proportion with a lactate clearance rate <10% (P=0.024). No significant associations were found with gender, BMI, cancer type, TNM stage, diabetes history, hypertension history, chemotherapy/radiotherapy history, first-attempt cannulation success rate, resuscitation success rate, infusion rate, urine output, or ALB levels before and after infusion (P>0.05) (Table 3).
Table 3.
Analysis of risk factors affecting LA clearance rate in patients
| Variable | Total | <10% (n=65) | ≥10% (n=453) | Statistic | P-Value |
|---|---|---|---|---|---|
| Age (years) | 65.00 [60.00, 70.00] | 65.00 [59.00, 70.00] | 69.00 [64.00, 78.00] | 5.345 | <0.001 |
| Gender | |||||
| Male | 384 (74.13%) | 337 (74.39%) | 47 (72.31%) | 0.129 | 0.720 |
| Female | 134 (25.87%) | 116 (25.61%) | 18 (27.69%) | ||
| BMI (kg/m2) | 24.20 [21.98, 25.99] | 24.21 [21.75, 26.00] | 24.12 [22.82, 25.85] | 0.543 | 0.587 |
| Cancer Type | |||||
| Gastric Cancer | 305 (58.88%) | 262 (57.84%) | 43 (66.15%) | 1.922 | 0.382 |
| Esophageal Cancer | 185 (35.71%) | 165 (36.42%) | 20 (30.77%) | ||
| Other | 28 (5.41%) | 26 (5.74%) | 2 (3.08%) | ||
| TNM Stage | |||||
| Stage II | 56 (10.81%) | 48 (10.60%) | 8 (12.31%) | 0.787 | 0.675 |
| Stage III | 134 (25.87%) | 120 (26.49%) | 14 (21.54%) | ||
| Stage IV | 328 (63.32%) | 285 (62.91%) | 43 (66.15%) | ||
| 24 h Blood Loss | |||||
| ≥1000 mL | 105 (20.27%) | 83 (18.32%) | 22 (33.85%) | 8.476 | 0.004 |
| <1000 mL | 413 (79.73%) | 370 (81.68%) | 43 (66.15%) | ||
| 24 h Blood Transfusion | |||||
| Yes | 281 (54.25%) | 244 (53.86%) | 37 (56.92%) | 0.214 | 0.643 |
| No | 237 (45.75%) | 209 (46.14%) | 28 (43.08%) | ||
| Time from Bleeding to Admission | |||||
| ≥12 h | 315 (60.81%) | 267 (58.94%) | 48 (73.85%) | 5.300 | 0.021 |
| <12 h | 203 (39.19%) | 186 (41.06%) | 17 (26.15%) | ||
| Hemorrhagic Shock | |||||
| Yes | 134 (25.87%) | 110 (24.28%) | 24 (36.92%) | 4.736 | 0.030 |
| No | 384 (74.13%) | 343 (75.72%) | 41 (63.08%) | ||
| Received Radiation/Chemotherapy | |||||
| Yes | 346 (66.80%) | 301 (66.45%) | 45 (69.23%) | 0.199 | 0.656 |
| No | 172 (33.20%) | 152 (33.55%) | 20 (30.77%) | ||
| Diabetes History | |||||
| Yes | 104 (20.08%) | 94 (20.75%) | 10 (15.38%) | 1.020 | 0.313 |
| No | 414 (79.92%) | 359 (79.25%) | 55 (84.62%) | ||
| Hypertension History | |||||
| Yes | 187 (36.10%) | 163 (35.98%) | 24 (36.92%) | 0.022 | 0.883 |
| No | 331 (63.90%) | 290 (64.02%) | 41 (63.08%) | ||
| Hemostasis Method | |||||
| Interventional Embolization | 333 (64.29%) | 45 (69.23%) | 288 (63.58%) | 0.792 | 0.374 |
| Endoscopic Hemostasis | 185 (35.71%) | 20 (30.77%) | 165 (36.42%) | ||
| Fluid Resuscitation Plan | |||||
| IV | 301 (58.11%) | 252 (55.63%) | 49 (75.38%) | 9.114 | 0.003 |
| IO | 217 (41.89%) | 201 (44.37%) | 16 (24.62%) | ||
| First Cannulation Success Rate | |||||
| Success | 444 (85.71%) | 389 (85.87%) | 55 (84.62%) | 0.073 | 0.787 |
| Failure | 74 (14.29%) | 64 (14.13%) | 10 (15.38%) | ||
| Resuscitation Success Rate | |||||
| Success | 483 (93.24%) | 424 (93.60%) | 59 (90.77%) | 0.722 | 0.395 |
| Failure | 35 (6.76%) | 29 (6.40%) | 6 (9.23%) | ||
| Time of Venous Access Establishment (min) | 4.50 [2.00, 6.00] | 4.00 [2.00, 6.00] | 5.00 [4.00, 6.00] | 1.327 | 0.185 |
| Time to Blood Pressure Recovery After Fluid Resuscitation (min) | 11.00 [9.00, 12.00] | 11.00 [9.00, 12.00] | 12.00 [10.00, 13.00] | 2.254 | 0.024 |
| Infusion Rate (mL/min) | 15.00 [13.00, 17.00] | 15.00 [13.00, 17.00] | 15.00 [13.00, 17.00] | 0.798 | 0.425 |
| Urine Output (mL) | 84.04±24.19 | 84.14±24.39 | 83.40±22.96 | 0.229 | 0.819 |
| Pre-fluid ALB (g/L) | 59.91±10.16 | 59.99±10.21 | 59.37±9.83 | 0.461 | 0.645 |
| 6 h Post-fluid ALB (g/L) | 126.00 [114.00, 138.00] | 127.00 [114.00, 139.00] | 121.00 [110.00, 133.00] | 1.651 | 0.099 |
Univariate and multivariate analysis of risk factors for lactate clearance
In univariate analysis, older age (OR=1.122, 95% CI: 1.081-1.168, P<0.001), blood loss ≥1000 mL (OR=0.438, 95% CI: 0.251-0.783, P=0.004), longer time from bleeding onset to admission (OR=0.508, 95% CI: 0.276-0.895, P=0.023), hemorrhagic shock (OR=0.548, 95% CI: 0.319-0.958, P=0.031), IO fluid resuscitation strategy (OR=0.409, 95% CI: 0.220-0.726, P=0.003), and longer time for blood pressure recovery (OR=1.148, 95% CI: 1.021-1.298, P=0.024) were associated with a lactate clearance rate <10%. Multivariate analysis identified age (OR=1.125, 95% CI: 1.081-1.175, P<0.001) and IO fluid resuscitation strategy (OR=0.289, 95% CI: 0.092-0.864, P=0.029) as independent risk factors for a lactate clearance rate <10%. Hemorrhagic shock showed a trend toward significance (OR=0.586, 95% CI: 0.324-1.076, P=0.080), while time for blood pressure recovery was not significant (OR=0.953, 95% CI: 0.754-1.201, P=0.682) (Figure 6).
Figure 6.
Univariate and multivariate analysis of risk factors affecting lactate clearance rate.
ROC curve analysis of risk predictive ability
A risk model was constructed: (Logit(p) = -4.416 + 0.118 × age + 0.765 × 24-hour blood loss ≥1000 mL + 0.816 × time from bleeding onset to admission + 1.241 × fluid resuscitation strategy). ROC curve analysis revealed that the risk model had the highest discriminatory ability for predicting a lactate clearance rate <10% (AUC=0.777), surpassing each variable alone [age (AUC=0.705), 24-hour blood loss (AUC=0.578), time from bleeding onset to admission (AUC=0.575), and fluid resuscitation strategy (AUC=0.599)]. Comparisons showed significant differences between the risk model and individual variables (P<0.05), indicating superior predictive performance of the risk model, followed by fluid resuscitation strategy (Figure 7A, 7B).
Figure 7.
ROC curve analysis of different markers. A. ROC curves for various markers and their combined prediction; B. Comparison of the AUC values among various markers and their combination. Note: POC: Receiver operating characteristic, AUC: Area under the curve.
Cox regression analysis of prognostic factors for 30-day all-cause mortality
Cox regression analysis identified age, diabetes history, and lactate clearance rate as independent prognostic factors for 30-day all-cause mortality. Each 1-year increase in age was associated with an increased mortality risk (univariate HR=0.384, 95% CI: 0.237-0.623, P<0.001; multivariate HR=0.572, 95% CI: 0.345-0.949, P=0.031). Diabetes history was linked to higher mortality risk (univariate HR=2.446, 95% CI: 1.179-5.074, P=0.016; multivariate HR=3.23, 95% CI: 1.507-6.926, P=0.003). Each unit increase in lactate clearance rate was associated with a reduced mortality risk (univariate HR=0.017, 95% CI: 0.009-0.031, P<0.001; multivariate HR=0.016, 95% CI: 0.009-0.031, P<0.001). While 24-hour blood loss was significant in univariate analysis (HR=0.544, 95% CI: 0.34-0.87, P=0.011), it was not significant in multivariate analysis (HR=0.831, 95% CI: 0.517-1.336, P=0.446) (Table 4; Figure 8).
Table 4.
Cox regression analysis of independent prognostic factors for 30-day all-cause mortality
| Variable | Univariate analysis | Multivariate analysis | ||||
|---|---|---|---|---|---|---|
|
|
|
|||||
| Beta | HR (95% CI) | P Value | Beta | HR (95% CI) | P Value | |
| Age | -0.956 | 0.384 (0.237-0.623) | <0.001 | -0.559 | 0.572 (0.345-0.949) | 0.031 |
| BMI | 0.599 | 1.82 (0.737-4.498) | 0.194 | |||
| Gender | ||||||
| Male | ||||||
| Female | 0.284 | 1.328 (0.834-2.115) | 0.232 | |||
| Cancer Type | ||||||
| Gastric Cancer | ||||||
| Esophageal Cancer | 0.088 | 1.092 (0.694-1.719) | 0.703 | |||
| Other | -0.111 | 0.895 (0.323-2.485) | 0.832 | |||
| TNM Stage | ||||||
| II | ||||||
| III | 0.013 | 1.014 (0.444-2.315) | 0.975 | |||
| IV | 0.195 | 1.215 (0.579-2.551) | 0.606 | |||
| 24 h Blood Loss | ||||||
| ≥1000 mL | ||||||
| <1000 mL | -0.610 | 0.544 (0.34-0.87) | 0.011 | -0.185 | 0.831 (0.517-1.336) | 0.446 |
| 24 h Blood Transfusion Treatment | ||||||
| Yes | ||||||
| No | -0.192 | 0.826 (0.532-1.281) | 0.393 | |||
| Time from Bleeding to Admission | ||||||
| ≥12 h | ||||||
| <12 h | -0.419 | 0.658 (0.411-1.052) | 0.081 | |||
| Hemorrhagic Shock | ||||||
| Yes | ||||||
| No | -0.021 | 0.979 (0.560-1.713) | 0.941 | |||
| Received Radiation/Chemotherapy | ||||||
| Yes | ||||||
| No | 0.086 | 1.089 (0.693-1.713) | 0.711 | |||
| Diabetes History | ||||||
| Yes | ||||||
| No | 0.895 | 2.446 (1.179-5.074) | 0.016 | 1.173 | 3.23 (1.507-6.926) | 0.003 |
| Hypertension History | ||||||
| Yes | ||||||
| No | 0.153 | 1.166 (0.735-1.847) | 0.515 | |||
| Fluid Therapy Plan | ||||||
| IV | ||||||
| IO | -0.519 | 0.595 (0.374-0.947) | 0.029 | 0.175 | 1.192 (0.734-1.935) | 0.479 |
| Hemostasis method | ||||||
| Interventional embolization | ||||||
| Endoscopic hemostasis | -0.463 | 0.630 (0.386-1.026) | 0.063 | |||
| First Intubation Success Rate | ||||||
| Success | ||||||
| Failure | 0.138 | 1.148 (0.635-2.077) | 0.648 | |||
| Resuscitation Success Rate | ||||||
| Success | ||||||
| Failure | -0.090 | 0.914 (0.37-2.258) | 0.845 | |||
| Time of Venous Access Establishment | ||||||
| ≥3 | ||||||
| <3 | -0.455 | 0.634 (0.376-1.07) | 0.088 | |||
| Post-Fluid BP Recovery | ||||||
| ≥10 | ||||||
| <10 | -0.292 | 0.747 (0.464-1.202) | 0.229 | |||
| Infusion Rate | ||||||
| ≥13 | ||||||
| <13 | -0.432 | 0.649 (0.344-1.225) | 0.182 | |||
| Urine Output | ||||||
| ≥105 | ||||||
| <105 | 0.291 | 1.337 (0.725-2.467) | 0.352 | |||
| Pre-Fluid ALB | ||||||
| ≥49 | ||||||
| <49 | 0.128 | 1.137 (0.602-2.146) | 0.692 | |||
| 6 h Post-Fluid ALB | ||||||
| ≥122 | ||||||
| <122 | 0.385 | 1.469 (0.953-2.265) | 0.082 | |||
| 6 h LA clearance rate | ||||||
| ≥0.16 | ||||||
| <0.16 | -4.101 | 0.017 (0.009-0.031) | <0.001 | -4.119 | 0.016 (0.009-0.031) | <0.001 |
Note: BMI: Body Mass Index, TNM: Tumor, Node, Metastasis, BP: Blood Pressure, ALB: Albumin, IV: Intravenous, IO: Intraosseous, IA: lactate.
Figure 8.
Kaplan-Meier survival curve analysis of factors affecting 30-day all-cause mortality. A. Effect of age on 30-day all-cause mortality rate; B. Effect of 24-hour blood transfusion treatment on 30-day all-cause mortality rate; C. Effect of diabetes history on 30-day all-cause mortality rate; D. Effect of fluid resuscitation plan on 30-day all-cause mortality rate; E. Effect of lactate clearance rate at 6 Hours on 30-day all-cause mortality rate.
Competing risk model for 30-day mortality risk
In the competing risk model, a 6-hour lactate clearance rate <0.16 was significantly associated with 30-day mortality risk in both univariate (HR=0.1, 95% CI: 0.047-0.235, P<0.001) and multivariate analyses (HR=0.11, 95% CI: 0.051-0.254, P<0.001). Other factors, including 24-hour blood loss, diabetes history, hemorrhagic shock, fluid resuscitation strategy, age, BMI, and 6-hour serum albumin levels, were not significant in multivariate analysis. Fluid resuscitation strategy (IO group) and gender also showed no significant impact on 30-day mortality risk (Table 5).
Table 5.
Application of competitive risk model in the assessment of 30-day mortality risk in patients with gastrointestinal bleeding
| Variable | Univariate analysis | Multivariate analysis | ||||
|---|---|---|---|---|---|---|
|
|
|
|||||
| Cof | HR (95% CI) | P Value | Cof | HR (95% CI) | P Value | |
| Gender | ||||||
| Male | ||||||
| Female | 0.307 | 1.360 (0.588-3.14) | 0.470 | |||
| Cancer Type | ||||||
| Gastric Cancer | ||||||
| Esophageal Cancer | -0.274 | 0.760 (0.311-1.858) | 0.550 | |||
| Other | 0.784 | 2.190 (0.651-7.373) | 0.210 | |||
| TNM Stage | ||||||
| II | ||||||
| III | -0.885 | 0.410 (0.12-1.414) | 0.160 | |||
| IV | -0.680 | 0.510 (0.186-1.379) | 0.180 | |||
| 24 h Blood Loss | ||||||
| ≥1000 mL | ||||||
| <1000 mL | -0.628 | 0.530 (0.231-1.231) | 0.140 | |||
| 24 h Blood Transfusion Treatment | ||||||
| Yes | ||||||
| No | -0.592 | 0.550 (0.239-1.282) | 0.170 | |||
| Time from Bleeding to Admission | ||||||
| ≥12 h | ||||||
| <12 h | -0.324 | 0.720 (0.313-1.671) | 0.450 | |||
| Hemorrhagic Shock | ||||||
| Yes | ||||||
| No | -0.493 | 0.610 (0.27-1.38) | 0.240 | |||
| Received Radiation/Chemotherapy | ||||||
| Yes | ||||||
| No | -0.057 | 0.940 (0.409-2.183) | 0.890 | |||
| Diabetes History | ||||||
| Yes | ||||||
| No | 0.620 | 1.860 (0.556-6.211) | 0.310 | |||
| Hypertension History | ||||||
| Yes | ||||||
| No | 0.002 | 1.000 (0.443-2.264) | 1.000 | |||
| Hemostasis method | ||||||
| Interventional embolization | ||||||
| Endoscopic hemostasis | -0.177 | 0.840 (0.363-1.933) | 0.680 | |||
| Fluid Therapy Plan | ||||||
| IV | ||||||
| IO | -0.452 | 0.64 (0.277-1.463) | 0.290 | |||
| First Intubation Success Rate | ||||||
| Success | ||||||
| Failure | -0.662 | 0.52 (0.122-2.185) | 0.370 | |||
| Resuscitation Success Rate | ||||||
| Success | ||||||
| Failure | 0.201 | 1.22 (0.285-5.247) | 0.790 | |||
| Age (year) | ||||||
| ≥74 | ||||||
| <74 | -0.111 | 0.9 (0.307-2.611) | 0.840 | |||
| BMI | ||||||
| ≥27.93 | ||||||
| <27.93 | 1.020 | 2.77 (0.374-20.523) | 0.320 | |||
| Time of Venous Access Establishment | ||||||
| ≥3 | ||||||
| <3 | -0.091 | 0.91 (0.384-2.167) | 0.840 | |||
| Post-Fluid BP Recovery | ||||||
| ≥10 | ||||||
| <10 | -0.118 | 0.89 (0.387-2.043) | 0.780 | |||
| Infusion Rate | ||||||
| ≥13 | ||||||
| <13 | -0.543 | 0.58 (0.174-1.946) | 0.380 | |||
| Urine Output | ||||||
| ≥105 | ||||||
| <105 | 0.514 | 1.67 (0.504-5.544) | 0.400 | |||
| Pre-Fluid ALB | ||||||
| ≥49 | ||||||
| <49 | 0.003 | 1 (0.301-3.342) | 1.000 | |||
| 6 h Post-Fluid ALB | ||||||
| ≥122 | ||||||
| <122 | 0.827 | 2.29 (1.031-5.065) | 0.042 | 0.541 | 1.72 (0.767-3.846) | 0.190 |
| 6 h LA clearance rate | ||||||
| ≥0.16 | ||||||
| <0.16 | -2.254 | 0.1 (0.047-0.235) | <0.001 | -2.172 | 0.11 (0.051-0.254) | <0.001 |
Note: BMI: Body Mass Index, TNM: Tumor, Node, Metastasis, ALB: Albumin, IV: Intravenous, IO: Intraosseous, BP: Blood Pressure, IA: lactate.
Mediating effects of diabetes history and 6-hour lactate clearance on fluid therapy and mortality
Causal mediation analysis evaluated the direct and indirect effects of fluid therapy on mortality, with diabetes history and 6-hour lactate clearance rate as potential mediators. The direct effect of diabetes history on mortality was significant (P=0.043), but its indirect effect was not (P=0.156). The 6-hour lactate clearance rate significantly mediated the effect of fluid therapy on mortality (indirect effect, P<0.001), indicating that fluid therapy influences mortality primarily through lactate clearance. The direct effect of the 6-hour lactate clearance rate approached significance (P=0.069) (Figure 9).
Figure 9.
Causal mediation analysis of fluid therapy’s effect on mortality through diabetes history and 6-hour lactic acid clearance rate. A. Mediating effect of diabetes history on the effect of fluid therapy on mortality; B. Mediating effect of 6-hour lactate clearance on the effect of fluid therapy on mortality. Note: ME: mediating effect, DE: direct effect.
Discussion
Gastrointestinal (GI) tumors, particularly gastric and esophageal cancers, are major contributors to global cancer-related mortality [15]. In advanced stages, patients often develop AGIB, a life-threatening complication driven by tumor invasion of blood vessels or ulcer formation. This can lead to massive hemorrhage, hemorrhagic shock, hypotension, organ failure, and death [16]. Intravenous (IV) fluid resuscitation is the standard approach for AGIB management, but challenges arise in cases of difficult vascular access or hemorrhagic shock, where slow IV infusion rates impede rapid volume restoration [17]. In recent years, IO fluid resuscitation has emerged as a promising alternative, enabling rapid establishment of a reliable infusion pathway and delivering fluids efficiently, offering significant advantages for critically ill patients, such as those with AGIB [18].
This retrospective study compared the efficacy of IO versus IV fluid resuscitation, focusing on 30-day mortality. The IO group demonstrated superior fluid resuscitation efficiency, with significantly higher first-attempt cannulation success rates, shorter times to establish infusion channels, and faster blood pressure recovery compared to the IV group. These findings align with prior research reporting an 88% cannulation success rate for IO in trauma patients [18] and an increasing use of IO as the initial vascular access method in shock patients [19]. IO’s independence from peripheral vascular conditions makes it particularly effective in cases of hemorrhagic shock or collapsed veins [20]. The anatomical advantage of intraosseous access facilitates rapid and reliable fluid delivery, swiftly improving hemodynamics [21], thereby enhancing treatment efficiency in AGIB patients. However, while fluid resuscitation efficiency differed significantly, its ultimate impact on mortality must be considered alongside hemostatic interventions. For instance, transcatheter arterial embolization (TAE), a common hemostatic approach for cancer-related upper GI bleeding, achieves a 99.1% technical success rate but has a 43.9% clinical failure rate due to rebleeding, contributing to a 17.8% 30-day mortality rate, with rebleeding directly causing 13.1% of deaths [13]. This underscores the critical role of synergistic “resuscitation-hemostasis” strategies. The lower mortality rate in the IO group (11.98% vs. 18.6% in the IV group) may reflect IO’s ability to provide a critical time window for subsequent hemostatic interventions. Delays in IV fluid delivery may postpone hemostasis, exacerbating outcomes. This hypothesis is supported by studies on septic shock, where early restrictive fluid resuscitation reduces complications such as acute respiratory distress syndrome (ARDS), enhances LA clearance, and shortens ICU stays, highlighting the importance of balancing fluid resuscitation efficiency with metabolic regulation [22].
The 30-day all-cause mortality rate was significantly lower in the IO group, likely due to rapid hemodynamic restoration, which reduced the risk of organ failure. Multivariate analysis identified IO fluid resuscitation as a significant predictor of higher LA clearance (HR=0.289), a critical marker of tissue perfusion and metabolic recovery. Effective LA clearance relies on adequate liver and kidney perfusion, which IO facilitates by delivering fluids directly into the bone marrow cavity, increasing cardiac output by 20%-30% [21]. In contrast, IV resuscitation may result in inadequate visceral perfusion due to peripheral vascular leakage, delaying LA clearance [19]. This mechanism partly explains the 3.5-fold increased mortality risk in patients with low LA clearance in this study.
Sepsis studies provide further evidence: LA clearance ≤40.3% [23] or lactate kinetic models (e.g., MELD-ΔLA score) [24] are strong predictor of 28-day mortality, while 20% albumin resuscitation in cirrhotic patients accelerates LA clearance and improves hemodynamics [25], underscoring liver metabolic capacity as a limiting factor. Additionally, baseline hemoglobin ≤60 g/L independently predicts mortality in TAE studies [13]. Low hemoglobin combined with low LA clearance creates a vicious cycle of oxygen delivery-metabolism imbalance, where insufficient hemoglobin limits oxygen transport, and lactate accumulation suppresses red blood cell production, forming a positive feedback loop.
Diabetes history was identified as an independent mortality risk factor, likely due to microvascular damage and mitochondrial dysfunction in diabetic patients. Hyperglycemia inhibits pyruvate dehydrogenase, impairing lactate conversion to the tricarboxylic acid cycle, thus delaying LA metabolism despite adequate resuscitation [26,27]. This effect is mirrored in septic patients, where the association between LA clearance and 28-day mortality is stronger with concurrent hyperbilirubinemia (TBIL≥2 mg/dL) [28], suggesting a synergistic lethal impact of liver dysfunction and metabolic disorders.
This study confirms IO’s advantages in GI tumor-related massive hemorrhage, improving fluid resuscitation efficiency, enhancing LA metabolism, and reducing mortality risk. IO serves as an effective alternative for patients with challenging venous access, such as those with peripheral vascular collapse or drug-induced vasoconstriction [20]. Its ability to rapidly establish access is critical in hemorrhagic shock [21], and its clinical adoption is increasing [19]. IO’s dual benefits-rapid restoration of effective circulating volume and preferential enhancement of visceral perfusion (e.g., liver and kidneys)-promote LA clearance and mitigate organ failure risk. However, reliance solely on fluid resuscitation has limitations, as evidenced by the 13.1% rebleeding-related mortality in TAE studies [13]. Future research should explore optimal timing for integrating IO with hemostatic therapies (e.g., endoscopy or embolization). For example, IO could stabilize hemodynamics during early resuscitation, creating a “golden time window” for TAE or surgical hemostasis. This strategy’s feasibility is supported by critical care studies: point-of-care echocardiography-guided septic shock management shortens LA clearance time by optimizing vasopressor use [29], and in polytrauma patients, a lactate/albumin ratio (LAR) ≥1.50 better predicts 28-day mortality than LA alone [30], suggesting that combined metabolic and inflammatory markers could enhance risk stratification. Additionally, the unique metabolic profiles of diabetic patients necessitate tailored approaches, such as combining IO resuscitation with enhanced glycemic control and antioxidant therapies (e.g., vitamin C) to improve LA metabolism and prognosis. Future studies could adapt sepsis prediction models (e.g., integrating IL-6, PCT, and LA clearance) [31] to develop AGIB-specific risk assessment tools for precision resuscitation and stratified interventions.
Mediation analysis provided critical insights into the mechanisms linking fluid resuscitation to mortality. The 6-hour LA clearance rate was a significant mediator of the relationship between fluid therapy and mortality, with IO resuscitation significantly enhancing LA clearance, a key indicator of tissue perfusion and metabolic recovery. The indirect effect of fluid therapy on mortality via LA clearance was highly significant, underscoring IO’s role in rapidly restoring hemodynamics, improving metabolic function, and reducing organ failure risk. In contrast, diabetes history was not a significant mediator, despite being a known mortality risk factor. This may reflect diabetes’ complex pathophysiology, including microvascular damage and mitochondrial dysfunction, which may impede metabolic recovery despite effective resuscitation. These findings highlight that while comorbidities like diabetes contribute to mortality risk, the primary mechanism by which fluid therapy improves survival is through efficient LA clearance and metabolic recovery, particularly with IO resuscitation. This emphasizes the need to understand how fluid resuscitation strategies influence key metabolic processes to optimize patient outcomes.
Despite robust evidence supporting IO use in GI tumor-related AGIB, this study has limitations. As a retrospective analysis, it is susceptible to selection bias, particularly in treatment allocation, and physician judgment may introduce confounding factors. Additionally, data from a single center may limit generalizability due to regional differences, necessitating validation through multicenter, large-scale studies. Patient heterogeneity, including variations in GI tumor types and clinical conditions, may also influence IO efficacy, warranting subgroup analyses in future research. Prospective, multicenter studies are needed to confirm IO’s efficacy across diverse populations and assess its impact on long-term survival, quality of life, and complications. Given the focus on LA clearance, future investigations should explore the role of LA clearance and other biomarkers in AGIB patients to develop personalized fluid resuscitation strategies. Furthermore, evaluating the combined effects of IO with other emergency interventions (e.g., endoscopy, embolization) could inform comprehensive treatment protocols for clinical practice.
Conclusion
This study demonstrates that intraosseous fluid resuscitation offers significant advantages in patients with gastrointestinal tumor-related massive hemorrhage. IO significantly enhances fluid resuscitation efficiency, improves lactate metabolism, and reduces 30-day mortality risk while maintaining a safety profile comparable to IV resuscitation.
Disclosure of conflict of interest
None.
References
- 1.Barrios P, Chawla A. Gastrointestinal malignancies: pancreatic cancer clinical trials in neoadjuvant chemotherapy. Cancer Treat Res. 2024;192:119–129. doi: 10.1007/978-3-031-61238-1_6. [DOI] [PubMed] [Google Scholar]
- 2.Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74:229–263. doi: 10.3322/caac.21834. [DOI] [PubMed] [Google Scholar]
- 3.Güven İE, Başpınar B, Durak MB, Yüksel İ. Comparison of urgent and early endoscopy for acute non-variceal upper gastrointestinal bleeding in high-risk patients. Gastroenterol Hepatol. 2023;46:178–184. doi: 10.1016/j.gastrohep.2022.05.002. [DOI] [PubMed] [Google Scholar]
- 4.Zhou M, Zhang J, Liu J, Smith SC Jr, Ma C, Ge J, Huo Y, Fonarow GC, Liu J, Hao Y, Gao F, Sun Y, Morgan L, Yang N, Hu G, Zeng Y, Han Y, Zhao D CCC-ACS Investigators. Proton pump inhibitors and in-hospital gastrointestinal bleeding in patients with acute coronary syndrome receiving dual antiplatelet therapy. Mayo Clin Proc. 2022;97:682–692. doi: 10.1016/j.mayocp.2021.11.037. [DOI] [PubMed] [Google Scholar]
- 5.Liu BX, Wang XZ, Yan YL, Xiao X, Yang L, Luo XF. Efficacy analysis of transcatheter arterial embolization in acute non-variceal upper gastrointestinal bleeding. Sichuan Da Xue Xue Bao Yi Xue Ban. 2022;53:398–403. doi: 10.12182/20220560203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Lee YJ, Min BR, Kim ES, Park KS, Cho KB, Jang BK, Chung WJ, Hwang JS, Jeon SW. Predictive factors of mortality within 30 days in patients with nonvariceal upper gastrointestinal bleeding. Korean J Intern Med. 2016;31:54–64. doi: 10.3904/kjim.2016.31.1.54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Chen MF, Lin Y, Chen LW. The prophylactic effect of proton pump inhibitors combined with enteral nutrition for preventing gastrointestinal hemorrhage after cardiovascular surgery in high-risk patients. Heart Surg Forum. 2019;22:E183–E190. doi: 10.1532/hsf.2279. [DOI] [PubMed] [Google Scholar]
- 8.Meynet G, Guechi Y, Habchi K, Schmutz T. Demystifying the intraosseous infusion technique. Rev Med Suisse. 2022;18:870–873. doi: 10.53738/REVMED.2022.18.779.870. [DOI] [PubMed] [Google Scholar]
- 9.Underwood C, Collins SN, van Eps AW, Mills PC, Allavena RE, Bailey SR, Medina Torres CE, Meizler A, Pollitt CC. Intraosseous infusion of the distal phalanx compared to systemic intravenous infusion for marimastat delivery to equine lamellar tissue. Vet J. 2015;205:357–363. doi: 10.1016/j.tvjl.2015.05.010. [DOI] [PubMed] [Google Scholar]
- 10.Weiss M, Wendel-Garcia PD, Grass B, Buehler PK, Kleine-Brueggeney M. Effect of vertical pump position on start-up fluid delivery of syringe pumps used for microinfusion. Paediatr Anaesth. 2023;33:1099–1107. doi: 10.1111/pan.14750. [DOI] [PubMed] [Google Scholar]
- 11.Johnson M, Inaba K, Byerly S, Falsgraf E, Lam L, Benjamin E, Strumwasser A, David JS, Demetriades D. Intraosseous infusion as a bridge to definitive access. Am Surg. 2016;82:876–880. [PubMed] [Google Scholar]
- 12.Bustamante S, Bajracharya GR, Cheruku S, Leung S, Mao G, Singh A, Mamoun N. Point-of-care ultrasound to identify landmarks of the proximal humerus: potential use for intraosseous vascular access. J Ultrasound Med. 2021;40:725–730. doi: 10.1002/jum.15442. [DOI] [PubMed] [Google Scholar]
- 13.Gong T, Tsauo J, Ding M, Jin L, Duan F, Yu Y, Li X. Transcatheter arterial embolization for cancer-related non-variceal upper gastrointestinal bleeding: a multicenter retrospective study of 107 patients. Diagn Interv Imaging. 2023;104:60–66. doi: 10.1016/j.diii.2022.07.006. [DOI] [PubMed] [Google Scholar]
- 14.Fischbach W, Klassert C, Schilling D, Sarah K, Schneider T, Cossmann T, Frieling T. Acute gastrointestinal bleeding: results from the prospective German electronic registry of nearly 11 000 endoscopies (ALGK GI-bleeding registry) Z Gastroenterol. 2021;59:446–453. doi: 10.1055/a-1404-3647. [DOI] [PubMed] [Google Scholar]
- 15.Li M, Li N, Piao H, Jin S, Wei H, Liu Q, Yu J, Wang W, Ma S, Jiang Y, Yao H, Shen Y, Fu J. Nanomaterials for targeted drug delivery for immunotherapy of digestive tract tumors. Front Immunol. 2025;16:1562766. doi: 10.3389/fimmu.2025.1562766. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Murakami T, Kamba E, Haga K, Akazawa Y, Ueyama H, Shibuya T, Hojo M, Nagahara A. Emergency endoscopic hemostasis for gastrointestinal bleeding using a self-assembling peptide: a case series. Medicina (Kaunas) 2023;59:931. doi: 10.3390/medicina59050931. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Wang Y, Zhou S, Wang L, Fang J, Zhang Y, Shi L, Lin G, Zhang M, Wang S. The use of intraosseous infusion in the early resuscitation of patients with extremely severe burns. J Burn Care Res. 2024;45:520–524. doi: 10.1093/jbcr/irad202. [DOI] [PubMed] [Google Scholar]
- 18.Rijnhout TWH, Kieft M, Klein WM, Tan ECTH. Effectiveness of intraosseous access during resuscitation: a retrospective cohort study. BMC Emerg Med. 2024;24:192. doi: 10.1186/s12873-024-01103-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Agostinucci JM, Alhéritière A, Metzger J, Nadiras P, Martineau L, Bertrand P, Gentilhomme A, Petrovic T, Adnet F, Lapostolle F. Evolution of the use of intraosseous vascular access in prehospital advanced cardiopulmonary resuscitation: the IOVA-CPR study. Int J Nurs Pract. 2024;30:e13244. doi: 10.1111/ijn.13244. [DOI] [PubMed] [Google Scholar]
- 20.Rittblat M, Tsur N, Karas A, Gendler S, Beer Z, Radomislensky I, Almog O, Tsur AM, Avital G, Talmy T. When to choose intraosseous access in prehospital trauma care: a registry-based study from the Israel defense forces. Chin J Traumatol. 2024 doi: 10.1016/j.cjtee.2024.08.008. [Epub ahead of print] [DOI] [PubMed] [Google Scholar]
- 21.Yang CH, Ng CJ, Huang HL, Chien LT, Wang MF, Chen CB, Tsai LH, Huang CH, Tseng HJ, Chien CY. Intraosseous and intravenous epinephrine administration routes in out-of-hospital cardiac arrest: survival and neurologic outcomes. J Am Heart Assoc. 2024;13:e036739. doi: 10.1161/JAHA.124.036739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hu X, Zhang J, Wang P, Dai X. Practice and effect evaluation of early restrictive fluid resuscitation strategy in the nursing care of patients with sepsis in the emergency department: a retrospective cohort study. JBI Evid Implement. 2023;21:269–276. doi: 10.1097/XEB.0000000000000365. [DOI] [PubMed] [Google Scholar]
- 23.Chaudhari M, Agarwal N. Study of significance of serum lactate kinetics in sepsis as mortality predictor. Indian J Crit Care Med. 2022;26:591–595. doi: 10.5005/jp-journals-10071-23935. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Li X, Gong M, Fu S, Zhang J, Wu S. Establishment of MELD-lactate clearance scoring system in predicting death risk of critically ill cirrhotic patients. BMC Gastroenterol. 2022;22:280. doi: 10.1186/s12876-022-02351-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Maiwall R, Kumar A, Pasupuleti SSR, Hidam AK, Tevethia H, Kumar G, Sahney A, Mitra LG, Sarin SK. A randomized-controlled trial comparing 20% albumin to plasmalyte in patients with cirrhosis and sepsis-induced hypotension [ALPS trial] J Hepatol. 2022;77:670–682. doi: 10.1016/j.jhep.2022.03.043. [DOI] [PubMed] [Google Scholar]
- 26.Zleczewski M, Alshoubi A. An unusual presentation of metformin-associated lactic acidosis: a case report. Cureus. 2025;17:e78870. doi: 10.7759/cureus.78870. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.See KC. Metformin-associated lactic acidosis: a mini review of pathophysiology, diagnosis and management in critically ill patients. World J Diabetes. 2024;15:1178–1186. doi: 10.4239/wjd.v15.i6.1178. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Takahashi N, Nakada TA, Walley KR, Russell JA. Significance of lactate clearance in septic shock patients with high bilirubin levels. Sci Rep. 2021;11:6313. doi: 10.1038/s41598-021-85700-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Killu K, Patino-Sutton C, Kysh L, Castriotta R, Oropello J, Huerta L, Engracia D, Merchant K, Wee CP, Cortessis VK. The association between integrating echocardiography use in the management of septic shock patients and outcomes in the intensive care unit: a systematic review and meta-analysis. J Ultrasound. 2024 doi: 10.1007/s40477-024-00958-w. [Epub ahead of print] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Arslan K, Sultan Sahin A. Lactate, lactate clearance, and lactate-to-albumin ratio in predicting mortality in patients with critical polytrauma: a retrospective observational study. Medicine (Baltimore) 2024;103:e40704. doi: 10.1097/MD.0000000000040704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Xie Y, Zhuang D, Chen H, Zou S, Chen W, Chen Y. 28-day sepsis mortality prediction model from combined serial interleukin-6, lactate, and procalcitonin measurements: a retrospective cohort study. Eur J Clin Microbiol Infect Dis. 2023;42:77–85. doi: 10.1007/s10096-022-04517-1. [DOI] [PMC free article] [PubMed] [Google Scholar]