Anticoagulation-free strategy within the first 24 h of extracorporeal cardiopulmonary resuscitation in refractory cardiac arrest patients: a retrospective cohort study

Abstract

Background

Extracorporeal cardiopulmonary resuscitation (ECPR), involving VA-ECMO during refractory cardiac arrest, has been linked to improved outcomes. While systemic anticoagulation is commonly used to prevent circuit thrombosis, bleeding and thrombotic complications remain frequent, and the need for early anticoagulation remains debated.

Methods

This single-center retrospective study included patients with refractory cardiac arrest who underwent ECPR between February 2020 and April 2025. Clinical data were extracted from medical records to assess the impact of an anticoagulation-free strategy during the first 24 h on bleeding, thrombosis, and outcomes.

Results

Patients were divided into two groups based on whether systemic anticoagulation was administered within the first 24 h following ECPR initiation: the non-anticoagulation group (n = 51) and the anticoagulation-free group (n = 50). Compared with the anticoagulation group, the anticoagulation-free group had a significantly lower incidence of bleeding events (72.5% vs. 50.0%, P = 0.021), while the incidence of thrombotic events did not differ significantly between the two groups (17.6% vs. 22.0%, P = 0.585). The anticoagulation-free group had the lowest demand for the rate of blood product use compare to anticoagulation group (64.0% vs. 80.4%, P = 0.044). Anticoagulation-free within the first 24 h, a higher SAVE score, and successful weaning from VA-ECMO were all independent predictors of 30-day survival.

Conclusion

The anticoagulation-free strategy within the first 24 h of ECPR in patients with refractory cardiac arrest was associated with a reduced risk of bleeding, without a significant increase in thrombotic events. This anticoagulation strategy may contribute to improved clinical outcomes in this high-risk population.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40001-025-03358-w.

Introduction

Refractory cardiac arrest, defined as failure to achieve return of spontaneous circulation (ROSC) after 15 min of cardiopulmonary resuscitation (CPR) or three defibrillation attempts [1, 2]. In patients without ROSC, survival rates are typically below 4% when hospital transport occurs during ongoing CPR [3].

Extracorporeal cardiopulmonary resuscitation (ECPR), the implementation of veno-arterial extracorporeal membrane oxygenation (VA-ECMO) during refractory cardiac arrest when ROSC is not achieved, has been shown in observational studies to improve survival and neurological outcomes [4–6]. However, results from randomized controlled trials (RCTs) have been inconsistent: the Advanced Reperfusion Strategies for Patients with OHCA and Refractory Ventricular Fibrillation Trial (ARREST) demonstrated a clear survival benefit, whereas the Prague OHCA and Early Initiation of Extracorporeal Life Support in Refractory OHCA Trial (INCEPTION) did not show significant improvements in long-term neurologically favorable survival [7–9]. These discrepancies likely reflect differences in patient selection, ECMO initiation time, and institutional expertise, underscoring the ongoing uncertainty regarding the generalizability of ECPR benefits.

In addition to these unresolved questions about efficacy, patients undergoing extracorporeal circulation face a substantial risk of both bleeding and thrombotic complications [10]. Such risks arise from a combination of factors, including the underlying pathological insult, blood–circuit interaction, coagulopathy, systemic inflammation, and the need for anticoagulation therapy [11].

Managing anticoagulation during extracorporeal circulation presents a persistent clinical challenge, as it requires maintaining a delicate balance between preventing thrombosis and minimizing bleeding risk [12]. Given this complexity, patients undergoing VA-ECMO, especially in the setting of ECPR, are particularly susceptible to bleeding and thrombotic events, including gastrointestinal bleeding, hemorrhagic stroke, circuit thrombosis, and ischemic stroke [13, 14]. Previous reports have indicated that significant risks remain associated with the use of extracorporeal circulation, particularly bleeding, intracranial hemorrhage, infection, and thrombosis [13, 15]. Bleeding is the most common hemocompatibility-related complication during extracorporeal circulation, accounting for 25–60% of all adverse events, and bleeding events are twice as common as thrombotic events [13, 16–18]. Despite improvements in extracorporeal circulation management over time, the incidence of major bleeding events has remained largely unchanged and is strongly associated with increased mortality.

In response to the substantial bleeding risk associated with systemic anticoagulation during ECMO, several strategies have been explored over the past decade [19]. Traditionally, unfractionated heparin (UFH) has been used to prevent circuit thrombosis, but this practice inevitably increases bleeding complications. To address this, modified anticoagulation regimens, such as dose reduction, individualized anticoagulation guided by coagulation monitoring, or the use of heparin-coated circuits, have been investigated as intermediate approaches. More recently, complete avoidance of systemic anticoagulation has emerged as a potential strategy, particularly in patients at high bleeding risk [20, 21]. For example, Olson et al. [20] reported a total incidence of thrombosis of 22.9% in a group of 201 patients supported with ECMO without systemic anticoagulation for a median of 4.75 days and a total duration of anticoagulant-free ECMO of 304.7 days. Importantly, these patients were managed using heparin- or other anticoagulant-bonded ECMO circuits. Thrombotic events were circuit-related in 13.4% (mainly oxygenator thrombosis requiring exchange) and patient-related in 9.5% of the cases, with a predominance of arterial thrombosis in VA-ECMO. This approach aims to minimize bleeding events while maintaining acceptable thrombotic risk through optimized circuit design and vigilant monitoring [22]. However, clinical evidence regarding this strategy remains limited and inconsistent, with variable reported outcomes that leave the safety and efficacy of extracorporeal circulation without systemic anticoagulation uncertain. Notably, the complete avoidance of systemic anticoagulation may potentially increase the risk of thrombotic complications, particularly in prolonged ECMO runs or in patients with underlying prothrombotic conditions [20].

Therefore, in the context of ECPR, we conducted this retrospective study to evaluate the clinical application of VA-ECMO support without systemic anticoagulation within the first 24 h in patients with refractory cardiac arrest, focusing on bleeding and thrombotic complications and their impact on clinical prognosis.

Methods

Study design and patient selection

This single-center, retrospective cohort study enrolled patients with refractory out-of-hospital cardiac arrest (OHCA) who received ECPR from February 2020 to April 2025 in Quzhou People’s Hospital. This study was approved by the ethics committees of Quzhou People’s Hospital (Approval number: 2020–05–06), and the requirement for informed consent was waived owing to the retrospective design and the use of deidentified data.

Patients were screened from the electronic medical records in Quzhou People’s Hospital from February 2020 to April 2025. A total of 121 consecutive patients with refractory cardiac arrest who received ECPR (defined as the initiation of VA-ECMO during ongoing CPR in patients with refractory cardiac arrest) were evaluated.

Inclusion criteria were as follows: (1) aged ≥ 18 years; (2) received ECPR; (3) diagnosed with refractory cardiac arrest, defined as failure to achieve return of spontaneous circulation (ROSC) after 15 min of cardiopulmonary resuscitation (CPR) or three defibrillation attempts [1, 2]; (4) survival for at least 24 h after VA-ECMO initiation. Exclusion criteria included: (1) age < 18 years; (2) incomplete clinical records (i.e., missing key data such as anticoagulation management, laboratory results, ECMO course, or 30-day outcomes); (3) patients with factors that would strongly influence the decision to use or withhold systemic anticoagulation, including active bleeding, severe thrombocytopenia, recent intracranial hemorrhage, and diagnosed hypercoagulable states; (4) VA-ECMO duration < 24 h. No anticoagulant or antiplatelet drugs were administered 24 h before VA-ECMO in all patients. Patients were divided into two groups based on whether systemic anticoagulation was administered within the first 24 h after ECPR initiation: the anticoagulation group and the anticoagulation-free group.

ECPR protocol

At our institution, once refractory cardiac arrest was recognized, the resuscitation team immediately evaluated patient eligibility for ECPR. When a patient was identified as a potential candidate, the ECMO team, consisting of experienced ICU physicians and perfusionists, was immediately activated. VA-ECMO cannulation was performed in the emergency department or intensive care unit under ongoing chest compressions. All ECMO circuits used were heparin-coated. The decision regarding anticoagulation strategy (anticoagulation-free vs. systemic anticoagulation) was made by the attending ICU physician based on the patient’s bleeding risk, hemodynamic status, and overall clinical condition. Post-cannulation management followed our institutional ECMO care guidelines, including hemodynamic support, targeted temperature management, and organ support as required.

ECMO implantation and management

The decision to initiate VA-ECMO following ECPR in patients with refractory cardiac arrest was made by experienced ICU physicians based on the patient’s clinical status, resuscitation duration, and neurological prognosis. Our ECMO circuits were constructed using Cortiva BioActive Surface tubing (Medtronic, Minneapolis, MN). Uncoated circuits were used only for patients who developed heparin-induced thrombocytopenia. VA-ECMO cannulation was performed percutaneously under ultrasound guidance using the Seldinger technique with puncture catheters (Maquet, Germany). The drainage cannula was placed in the femoral vein, and returned into the contralateral femoral artery. A 6 F arterial sheath was routinely inserted into the superficial femoral artery for distal limb perfusion. The membrane oxygenator kits were provided by Xenios AG (Germany) or Maquet (Germany).

Anticoagulation strategy

In our center, anticoagulation management during VA-ECMO was performed with UFH infusion. The target activated partial thromboplastin time (aPTT) was maintained at 50–70 s, and platelet counts were kept above 50 × 10^9/L with transfusion support if necessary. Laboratory monitoring included aPTT, International normalized ratio (INR), fibrinogen, D-dimer, platelet count, and hemoglobin, which were routinely assessed at 0, 1, 2, 4, 8, 12, and 24 h after VA-ECMO initiation to assess the risk of bleeding and thrombosis. The decision to initiate or withhold anticoagulation within the first 24 h after ECPR was made at the discretion of the treating physician, based on bleeding risk, hemodynamic status, and laboratory results. In patients who experienced bleeding complications, aPTT was generally maintained within this target range, with individualized adjustments as clinically indicated.

Study outcomes

The primary outcomes included:

1. Bleeding complications: were defined according to the ELSO criteria. Major bleeding was defined as clinically overt bleeding associated with hemodynamic instability, requiring surgical intervention, or necessitating transfusion of ≥ 2 units of red blood cells within 24 h. Minor bleeding was defined as any bleeding event not meeting these criteria [23].

2. Thrombotic complications: included oxygenator clotting requiring circuit change, deep vein thrombosis confirmed by ultrasound, pulmonary embolism, ischemic stroke, or other clinically relevant thromboembolic events diagnosed by imaging or autopsy.

3. 30-day survival: defined as survival at 30 days after ECPR initiation, irrespective of ICU or hospital discharge status.

Secondary outcomes included:

1. Dynamic changes in coagulation parameters within the first 24 h after ECMO initiation.

2. Blood product utilization.

3. ECMO-related process indicators, such as duration of mechanical ventilation, duration of VA-ECMO support, and successful VA-ECMO weaning.

Risk factors for bleeding and thrombotic events were evaluated using univariable and multivariable logistic regression analyses. Variables considered clinically relevant or previously reported as potential risk factors (e.g., age, sex, obesity, use of antiplatelet therapy, Length of VA-ECMO, and d Length of MV) were entered into the multivariable models irrespective of their significance in univariable analyses [16, 24, 25].

Data collection

Clinical data were retrospectively extracted from the electronic medical records of Quzhou People’s Hospital. Data collected included patients’ demographics, medical history, ECPR indications, VA-ECMO parameters, anticoagulation strategy, laboratory test results, bleeding and thrombotic events, complications, SAVE score [26] (Survival After VA-ECMO, a validated tool to predict survival in patients with refractory cardiogenic shock on VA-ECMO, based on clinical, laboratory, and hemodynamic variables such as age, underlying cardiac disease, lactate, and platelet count) and clinical outcomes. Data quality was ensured by cross-checking records by two independent investigators.

Statistical analysis

Data were analyzed using SPSS software (version 26; IBM Corp., Armonk, NY, USA) and R software (version 4.4.1; R Foundation for Statistical Computing, Vienna, Austria). Graphs and figures were generated with GraphPad Prism (version 10; GraphPad Software, San Diego, CA, USA) and OriginPro (version 2024b; OriginLab Corp., Northampton, MA, USA). Continuous variables were tested for normality using the Kolmogorov–Smirnov test and expressed as mean ± standard deviation (SD) for normally distributed data or median and interquartile range (IQR) for non-normally distributed data. Comparisons between groups were performed using the independent samples t-test or Mann–Whitney U test, as appropriate. Categorical variables, including the incidence rates of bleeding and thrombotic events, were compared using the chi-square test or Fisher’s exact test when expected cell counts were small. Line graphs with error bars were used to illustrate the temporal trends of coagulation indicators between groups at different time points. Forest plot was constructed to visualize the results of univariate and multivariate logistic regression models evaluating factors associated with 30-day outcomes. Survival analysis was conducted using the Kaplan–Meier method and compared by the log-rank test. A two-sided P value < 0.05 was considered statistically significant.

Results

Clinical characteristics of patients

From February 2020 to April 2025, a total of 121 patients with refractory cardiac arrest who received ECPR were enrolled in this study. After excluding 1 patient under the age of 18, 6 patients with incomplete clinical records, 4 patients with contraindications to anticoagulation, and 9 patients with a VA-ECMO duration of less than 24 h, 101 patients were finally included in the analysis. Based on whether systemic anticoagulation was administered within the first 24 h after ECPR initiation, patients were divided into the anticoagulation group (n = 51) and the anticoagulation-free group (n = 50) (Fig. 1).

Fig. 1.

The study flowchart. ECPR extracorporeal cardiopulmonary resuscitation; VA-ECMO, veno-arterial extracorporeal membrane oxygenation

The two groups were generally comparable in baseline characteristics (Table 1). Median age was similar between the anticoagulation and anticoagulation-free groups [56 (47–66) vs. 61 (48.5–68.75), P = 0.343], and the proportion of male patients was not significantly different (82.4% vs. 68.0%, P = 0.096). No significant differences were found in body mass index, comorbidities, or initial laboratory parameters. Severity scores, including APACHE II, SOFA, and SAVE scores, were also comparable between groups.

Table 1.

Patient demographic data and clinical characteristics

Variable	Anticoagulation (n = 51)	Anticoagulation-free (n = 50)	P value
Age	56 (47–66)	61 (48.5–68.75)	0.343
Sex, n (%)
Male	42 (82.4)	34 (68.0)	0.096
Female	9 (17.6)	16 (32.0)
Body mass index	24.0 (22.0–26.0)	24.0 (21.8–27.0)	0.782
Primary cause of cardiac arrest, n (%)
ACS	7 (13.73)	6 (12.00)	0.797
Malignant arrhythmia	4 (7.84)	6 (12.00)	0.487
Myocarditis	7 (13.73)	5 (10.00)	0.565
Poisoning/drug overdose	6 (11.76)	3 (6.00)	0.312
Drowning/Electric shock	2 (3.92)	1 (2.00)	0.571
Non-specific cardiogenic shock	25 (49.02)	29 (58.00)	0.777
Comorbidities, n (%)
Hypertension	12 (23.5)	13 (26.0)	0.775
Coronary artery disease	9 (17.6)	11 (22.0)	0.585
Renal dysfunction	14 (27.5)	16 (32.0)	0.619
Diabetes	7 (13.7)	11 (22.0)	0.280
Initial VA-ECMO circuit setting
Flow rate (L/min)	2.7 (2.2–3.0)	2.5 (2.0–2.8)	0.137
Speed (rpm)	2450 (2100–3000)	2400 (2190–2700)	0.479
Initial laboratory test
Lactate (mmol/L)	12.6 (5.4–20.0)	11.9 (6.55–22.0)	0.390
Platelets (× 109/L)	167 (122–196)	166 (120–216.5)	0.747
Hemoglobin (g/dL)	133 (110–152)	127.5 (101.8–140.3)	0.254
Total bilirubin (mg/dL)	10.4 (7.8–24.7)	13.1(7.9–22.5)	0.644
Creatinine (µmol/L)	115 (90.0–153.0)	116.85 (93.6–143.4)	0.757
Blood urea nitrogen (mmol/L)	6.9(4.6–10.9)	7.5 (5.1–9.5)	0.919
White blood count (× 109/L)	13.2 (8.5–16.2)	14.5 (10.5–20.5)	0.162
C-reactive protein (mg/L)	5.1 (2.27–35.4)	7.2 (3.5–31.3)	0.340
Severity and prognostic score
APACHE-II score	26 (23–29)	26 (21–30)	0.822
SOFA score	19 (16–20)	19 (17–20)	0.674
SAVE score	−5 (− 10 to − 2)	− 6 (− 9–0)	0.600
Use of antiplatelet therapy, n (%)	10 (19.61)	6 (12.00)	0.298
ICU treatment
CRRT, n (%)	41 (80.4)	44 (88.0)	0.647
IABP, n (%)	9 (17.7)	8 (16.0)	0.841
Length of MV (days)	2.2(1.3–6.7)	5.3 (1.7–10.2)	0.039
Length of VA-ECMO (days)	1.7(1.2–3.7)	3.2 (1.6–6.5)	0.012
VA-ECMO weaning, n (%)	12 (23.5)	21 (42.00)	0.049
Clinical outcomes, n (%)
30-day survival	9 (17.6)	22 (44.0)	0.004

Data are presented as the median (25th, 75th percentiles) or number (%).

Non-specific cardiogenic shock refers to severe heart failure not attributable to ACS, myocarditis, or other defined etiologies.

APACHE-II acute physiology and chronic health evaluation II, SOFA sequential organ failure assessment, SAVE survival after veno-arterial-ECMO, VA-ECMO veno-arterial extracorporeal membrane oxygenation, MV mechanical ventilation, CRRT continuous renal replacement therapy, IABP intra-aortic balloon pump, ACS acute coronary syndrome.

Patients in the anticoagulation-free group had a significantly longer duration of mechanical ventilation (MV) [(5.3 (1.7–10.2) vs. 2.2 (1.3–6.7), P = 0.039] and VA-ECMO support [3.2 (1.6–6.5) vs. 1.7 (1.2–3.7), P = 0.012] compared to the anticoagulation group. Additionally, the anticoagulation-free group also demonstrated a higher rate of successful VA-ECMO weaning (42.0% vs. 23.5%, P = 0.049) and a significantly higher rate of 30-day survival (44.0 vs. 17.6, P = 0.004).

Dynamic changes in coagulation indicators

To evaluate the impact of anticoagulation-free management within the first 24 h on the coagulation system, coagulation parameters were continuously monitored from baseline (0 h) to 24 h after ECMO initiation (Fig. 2).

Fig. 2.

Dynamic changes in coagulation indicators within 24 h after VA-ECMO initiation. *P < 0.05 anticoagulation versus anticoagulation-free group

Coagulation parameters showed dynamic changes following ECMO initiation. aPTT levels rose rapidly in both groups, peaking at 4 h and remaining elevated at 24 h, with consistently higher values in the anticoagulation group. INR level gradually increased, peaking at 12 h, followed by a slight decline, while remaining higher in the anticoagulation group. Fibrinogen levels declined significantly post-initiation, more markedly in the anticoagulation group, with partial recovery by 24 h but not to baseline. D-dimer levels progressively decreased, with higher values observed in the anticoagulation-free group throughout. Platelet and hemoglobin levels declined early and subsequently stabilized, showing similar trends between groups.

Bleeding, thrombotic events and blood product use

The anatomical distribution of bleeding and thrombotic events is illustrated in Fig. 3. Compared with anticoagulation and anticoagulation-free groups, the anticoagulation group had the more incidence of bleeding events (72.5 vs. 50.0, P = 0.021) compared to the anticoagulation-free group (Table 2). Among specific bleeding sites, no statistically significant differences were observed, though gastric and intestinal bleeding were the most common in both groups.

Fig. 3.

The anatomical distribution of bleeding and thrombotic events

Table 2.

Bleeding, thrombotic events and blood product use

Variable	Anticoagulation (n = 51)	Anticoagulation-free (n = 50)	P value
Bleeding events (within 24 h), n (%)	37 (72.5)	25 (50.0)	0.021
Cerebral	4 (7.8)	3 (6.0)	0.717
Ear/nose/throat	2 (3.9)	6 (12.0)	0.135
Gastric	20 (39.2)	17 (34.0)	0.588
Intestinal	7 (13.7)	4 (8.0)	0.358
Tracheal/pulmonary	6 (11.8)	3 (6.0)	0.312
Site of cannulation	5 (9.8)	7 (14.0)	0.517
Thrombotic events, n (%)	9 (17.6)	11 (22.0)	0.585
Jugular vein	1 (2.0)	5 (10.0)	0.089
Deep lower extremity veins	6 (11.8)	5 (10.0)	0.777
Femoral artery	1 (2.0)	1 (2.0)	0.989
DIC, n (%)	8 (15.7)	4 (8.0)	0.235
Blood production use, n (%)	41 (80.4)	32 (64.0)	0.044
Packed red blood cells	34(66.7)	21(42.0)	0.013
Platelets	10(19.6)	6(12.0)	0.298
Plasma	39(76.5)	28(56.0)	0.030
Cryoprecipitate	9(17.6)	12(24.0)	0.434

Data are presented as the number (%). DIC disseminated intravascular coagulation

Risk factors for bleeding events were evaluated using logistic regression. In the univariable analysis, the absence of anticoagulation within 24 h was associated with a significantly lower risk of bleeding events [0.348 (0.165–0.866), P = 0.021]. In addition, longer duration of mechanical ventilation was significantly associated with a reduced risk of bleeding [0.938 (0.886–0.994), P = 0.029]. Other variables did not show statistically significant associations (P > 0.05). In the multivariable model, after adjusting for clinically relevant factors, the anticoagulation-free within 24 h strategy remained independently associated with a lower risk of bleeding events [0.386 (0.156–0.956), P = 0.040], whereas the effects of mechanical ventilation duration and antiplatelet therapy were not statistically significant (Table 3).

Table 3.

Risk factors for bleeding events

Variables	Univariable OR (95% CI)	P. value	Multivariable OR (95% CI)	P. value
Age	1.005 (0.979–1.031)	0.716	1.004 (0.976–1.033)	0.782
Sex	1.675 (0.671–4.180)	0.269	1.490 (0.537–4.132)	0.443
Body mass index	1.090 (0.950–1.349)	0.218
Anticoagulation-free within 24 h	0.348 (0.165–0.866)	0.021	0.386 (0.156–0.956)	0.040
Length of MV	0.938 (0.886–0.994)	0.029	0.944 (0.877–1.016)	0.122
Length of VA-ECMO	0.897 (0.788–1.020)	0.097	1.006 (0.850–1.190)	0.948
Use of antiplatelet therapy	1.131 (0.431–2.964)	0.803	1.672 (0.568–4.921)	0.351
CRRT	0.776 (0.263–2.288)	0.646
IABP	1.116 (0.386–3.227)	0.840

VA-ECMO veno-arterial extracorporeal membrane oxygenation, MV mechanical ventilation, CRRT continuous renal replacement therapy, IABP intra-aortic balloon pump

Data are presented as the number of events (percentage) in each group: anticoagulation group/anticoagulation-free group.

Thrombotic events occurred at similar rates between groups (17.6% vs. 22.0%, P = 0.585), with a trend toward more jugular vein thrombosis in the anticoagulation-free group (10.0% vs. 2.0%, P = 0.089), though not statistically significant (Table 2). Logistic regression was also used to analyze risk factors for thrombotic events (Table 4). In the univariable analysis, both older age and longer duration of mechanical ventilation were significantly associated with an increased risk of thrombosis. In the multivariable model, after adjusting for clinically relevant factors, older age remained independently associated with thrombotic risk [1.079 (1.012–1.149), P = 0.019]. Mechanical ventilation duration showed only a borderline association [1.080 (0.999–1.168), P = 0.052]. Notably, the anticoagulation-free strategy within 24 h was not associated with an increased risk of thrombotic events [2.156 (0.495–9.387), P = 0.306], suggesting that anticoagulation-free within 24 h strategy during the early phase of ECPR may be a feasible and safe approach.

Table 4.

Risk factors for thrombotic events

Variables	Univariable. OR (95% CI)	P. value	Multivariable. OR (95% CI)	P. value
Age	1.053 (1.002–1.107)	0.042	1.079 (1.012–1.149)	0.019
Sex	0.985 (0.245–3.965)	0.983	0.918 (0.175–4.807)	0.919
Body mass index	1.006 (0.823–1.229)	0.955	0.939 (0.744–1.185)	0.594
Anticoagulation-free within 24 h	0.978 (0.293–3.264)	0.971	2.156 (0.495–9.387)	0.306
Length of MV	1.069 (1.008–1.134)	0.025	1.080 (0.999–1.168)	0.052
Length of VA-ECMO	1.138 (0.962–1.347)	0.133	1.080 (0.893–1.381)	0.344
Platelet transfusion	1.354 (0.373–4.911)	0.645
CRRT	0.448 (0.054–3.740)	0.459
IABP	0.912 (0.179–4.654)	0.912

VA-ECMO veno-arterial extracorporeal membrane oxygenation, MV mechanical ventilation, CRRT continuous renal replacement therapy, IABP intra-aortic balloon pump

Overall, 73.3% of patients received allogeneic blood transfusions (Table 2). The anticoagulation group had a higher rate of blood product use than the anticoagulation-free group (80.4 vs. 64.0, P = 0.044). Notably, the use of packed red blood cells and plasma was more frequent in the anticoagulation group. Platelet and cryoprecipitate transfusions were less common and showed no marked differences between groups.

Prognostic factors of 30-day survival

Predictive factors for 30-day survival were evaluated, including age, sex, anticoagulation-free within 24 h, SAVE score, VA-ECMO weaning, bleeding events, and thrombotic events. Univariable analysis showed that anticoagulation-free within 24 h, higher SAVE score, and successful VA-ECMO weaning were associated with improved 30-day survival (Fig. 4). In multivariable models, anticoagulation-free [2.879 (1.096–7.568), P = 0.032], SAVE score [1.116 (1.008–1.237), P = 0.035], and VA-ECMO weaning [2.962 (1.130–7.769) P = 0.027] remained significant prognostic indicators.

Fig. 4.

Logistic regression results and forest plot for 30-day survival. Note: the results from univariable and multivariable logistic regression are presented in black and red font. The x-axis represents odds ratios on a log₁₀ scale

Kaplan–Meier curves based on these predictors showed that patients who received anticoagulation-free within 24 h, higher SAVE score (cut-off value ≥ − 7) or were successfully weaned from VA-ECMO had a higher probability of 30-day survival (Fig. 5).

Fig. 5.

Kaplan–Meier curves for 30-day survival based on different prognostic factors. (a) Patients who received an anticoagulation-free strategy within 24 h or systemic anticoagulation; (b) VA-ECMO weaning status: successful or failure; (c) SAVE score ≥ −7 or < −7

Discussion

ECPR is increasingly recognized as an advanced resuscitative strategy that employs VA-ECMO during conventional CPR to provide hemodynamic support and facilitate targeted interventions in patients with refractory cardiac arrest [27]. However, numerous studies have demonstrated that bleeding and thrombotic events have become major complications associated with extracorporeal life support, particularly among patients undergoing ECPR [13, 14, 27, 28].

Bleeding and thrombotic events represent the most frequent and clinically significant complications in patients undergoing ECPR and ECMO support. A comprehensive review reported that bleeding, most commonly at cannulation sites and including systemic hemorrhages, occurs in 30–70% of ECPR patients, making it the predominant adverse event associated with this modality [28–30]. Large-scale registry data from the Extracorporeal Life Support Organization (ELSO), including nearly 12,000 patients, further confirmed that hemocompatibility-related adverse events are common, with 62.1% being bleeding events. The most fatal complications included intracranial hemorrhage (OR: 7.71), pulmonary hemorrhage (OR: 3.08), gastrointestinal hemorrhage (OR: 1.95), and ischemic stroke (OR: 2.31), all of which were independently associated with increased mortality [13]. Similarly, observational studies have reported that 33% of patients experienced bleeding, 10% venous thrombosis, 8% central nervous system complications, and 7% gastrointestinal bleeding during ECMO support [15]. Taken together, these findings underscore that bleeding events occur approximately twice as often as thrombotic events, with intracranial and gastrointestinal bleeding, as well as ischemic stroke, representing the most devastating complications. These findings underscore the importance of anticoagulation strategies in balancing bleeding and thrombotic risks, which may ultimately affect patient outcomes [10]. Therefore, achieving an optimal balance between thrombosis prevention and bleeding risk remains a persistent challenge in the anticoagulation management of critically ill patients during extracorporeal circulation [12].

UFH remains the standard anticoagulant in extracorporeal circulation due to its short half-life and reversibility. However, anti-Xa monitoring lacks correlation with actual clot formation, and the risk of heparin-induced thrombocytopenia further complicates its use [31]. Heparin resistance, influenced by hypercoagulability or thrombocytosis, further complicates management [32]. Alternatives like argatroban and bivalirudin are used, especially in renal or heparin-compromised patients, but present their own limitations in dosing and monitoring [33, 34].

These challenges have prompted interest in anticoagulation-free strategies, particularly in patients at high bleeding risk or with newer circuit designs, although evidence remains limited and further investigation is warranted. With advances in technology, particularly the use of heparin-coated circuits, ECMO may potentially be conducted with minimal or even no systemic anticoagulation in selected cases [20, 21, 35, 36]. A systematic review of adult ECMO cases (1977–2019) suggested that anticoagulation-free ECMO may be safe and feasible in selected high-bleeding-risk patients. Reported rates of major bleeding (5.8%), circuit thrombosis (9.5%), and patient thrombosis (3.9%) were comparable to or lower than those in anticoagulated ECMO cohorts (≥ 16%, 10%, and 18%, respectively). Although most studies were retrospective and small-scale, these findings may indicate the potential of low- or no-anticoagulation strategies under specific conditions [36, 37].

The feasibility of anticoagulation-free ECMO in selected patients may be explained by several factors. Anticoagulant-bonded circuits reduce thrombogenicity, while optimized flow dynamics—maintaining adequate pump speed and minimizing stasis—further decrease thrombus formation. Patient-specific coagulation profiles, including normal platelet function and absence of hypercoagulable states, also influence thrombotic risk. These factors together likely contribute to the safety of short-term anticoagulation-free strategies in certain ECPR patients, though prospective studies are needed for validation. However, most of the existing studies were limited by small sample sizes, and the overall level of clinical evidence remains low. Moreover, the risk of thrombotic events with complete anticoagulation withdrawal cannot be ignored. Therefore, building on these insights, we first conducted a retrospective study to evaluate the feasibility and safety of a 24-h anticoagulation-free strategy in patients undergoing ECPR, providing preliminary evidence to inform future prospective research.

Our study suggests that anticoagulation-free strategy during the first 24 h after VA-ECMO initiation in ECPR patients may be associated with a reduced risk of bleeding without a significant increase in thrombotic events, and could potentially improve short-term outcomes. The anticoagulation-free group experienced a significantly lower incidence of bleeding events (72.5% vs. 50.0%, P = 0.021), while the incidence of thrombotic events did not differ significantly between the two groups (17.6% vs. 22.0%, P = 0.585) (Table 2), which is consistent with findings from previous studies [37]. The most common bleeding sites in both groups were gastric, cannula site, and intestinal (Fig. 3). Notably, aPTT values showed only a modest difference between the anticoagulation and no-anticoagulation groups (Fig. 2). This finding may be partly explained by the clinical characteristics of ECPR patients, as most deaths or ECMO withdrawals occurred within the first 24 h, resulting in incomplete aPTT data beyond this period. In addition, clinicians tended to adopt a conservative anticoagulation target during the early phase of ECMO to minimize bleeding risk, which could have further narrowed the intergroup difference. Moreover, as previously reported, aPTT is a variable and sometimes unreliable marker in ECMO patients, with no consistently defined threshold for bleeding risk [38, 39]. Another factor that may have contributed to the modest intergroup differences is the universal use of heparin-coated ECMO circuits in our cohort. Heparin-coated surfaces can reduce blood–surface activation of coagulation and inflammation, often allowing lower systemic anticoagulation targets or intermittent anticoagulation in clinical practice [40]. Consequently, the use of heparin-coated circuits in both groups might have attenuated differences in systemic coagulation parameters such as aPTT, acting as a shared intervention that confounded the effect of systemic heparin therapy. Collectively, these factors may account for the similar aPTT profiles observed and suggest that short-term anticoagulation-free management may not markedly alter systemic coagulation monitoring results.

Thrombotic events most frequently occurred in the deep lower extremity veins, followed by the jugular vein and the femoral artery. Further analysis demonstrated that being anticoagulation-free within 24 h was independently associated with a reduced risk of bleeding events [0.348 (0.165–0.866), P = 0.021 in univariate analysis; 0.386 (0.156–0.956), P = 0.040 in multivariate analysis] (Table 3). A forest plot was used to visualize the logistic regression analysis of 30-day outcomes (Fig. 4). The results demonstrated that anticoagulation-free within 24 h, along with the SAVE score and successful VA-ECMO weaning, were independent predictors of 30-day prognosis. Subsequent Kaplan–Meier survival analysis further confirmed the positive impact of all three factors on 30-day survival, as patients with these characteristics exhibited significantly improved outcomes (Fig. 5). We further analyzed the interaction between anticoagulation status, SAVE score, and VA-ECMO weaning. The results indicated that SAVE score did not significantly influence anticoagulation decisions (F = 0.744, P = 0.767), whereas VA-ECMO weaning was significantly associated with anticoagulation status (F = 5.937, P = 0.018). Importantly, the interaction between SAVE score and weaning success was not significant (F = 0.490, P = 0.923) (Supplementary File 1). These findings support the robustness of the anticoagulation-free approach, suggesting that the effect of the 24-h anticoagulation-free strategy on outcomes is consistent across different SAVE scores and independent of VA-ECMO weaning.

Limitation

This study has several limitations. It was a single-center, retrospective study with a relatively small sample size, which may limit generalizability and statistical power. Potential confounding factors could not be fully controlled. The study was not adequately powered to detect differences in rare thrombotic events, and long-term outcomes were not assessed. In addition, due to the high rate of early mortality and ECMO withdrawal among ECPR patients, laboratory data beyond the initial 24 h were not consistently available. This limitation may have affected our ability to evaluate temporal trends in coagulation parameters such as aPTT between the anticoagulation and no-anticoagulation groups. Moreover, patients who died within 24 h after ECMO initiation were excluded, which may underestimate early mortality. Therefore, the impact of the 24-h anticoagulation-free strategy on thrombotic complications should be interpreted with caution. Larger multicenter prospective studies are needed to validate these findings.

Conclusion

In this study, we found that anticoagulation-free strategy during the first 24 h of ECPR in patients with refractory cardiac arrest could significantly reduce the incidence of bleeding events, without increasing the risk of thrombotic events. Moreover, patients managed with this strategy was associated with improved 30-day survival. These findings suggest that an early anticoagulation-free strategy may be a safe and effective approach in refractory cardiac arrest patients undergoing ECPR.

Author contributions

Zhao. JJ: conceptualization, methodology, software, formal analysis, resources, data curation, writing-original draft and supervision. Tang. J: resources and data curation. Wang. KY: resources. Wang. DQ: resources. Fang. HL: resources. Luo. J: conceptualization, resources, data curation, supervision.

Funding

This work was supported in part by grants from the Zhejiang Provincial Medical and Health Science and Technology Program (No. 2021KY1192) and Project of Zhejiang Provincial Department of Health (No. 2023KY1296).

Data availability

Data are available from corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

These retrospective studies were approved by the ethics committees of Quzhou People’s Hospital (Approval number: 2020–05-06). The studies were conducted in accordance with local legislation, institutional requirements at each site, and the Declaration of Helsinki. Informed consent was obtained from all participants and/or their legal guardians.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.