Correlation of anticoagulation assays with bleeding during extracorporeal membrane oxygenation at a tertiary pediatric hospital

Abstract

Introduction

Due to anticoagulation during Extracorporeal Membrane Oxygenation (ECMO), patients face a higher risk of bleeding. This study aimed to assess the correlation between anticoagulation assays and hemorrhage in ECMO patients.

Methods

We reviewed records from patients in our critical care units, comparing anticoagulation parameters with clinical outcomes. Statistical analyses included the Wilcoxon rank sum test for continuous variables and chi-square or Fisher’s exact tests for categorical variables. We used Kaplan-Meier curves to depict hemorrhage incidence and logistic regression to identify factors associated with bleeding and mortality. Linear regression assessed coagulation factors linked to length of stay (LOS).

Results

Of 76 patients, 32 (42.1%) experienced hemorrhage; 15 (46.9%) of these survived. Survivors had a higher proportion without bleeding (68.2%). Hemorrhage patients had more circuit changes (18.8% vs 2.3%, p = .01). The 10-, 20-, and 30-days cumulative hemorrhage incidences were 36.9%, 58.8%, and 72.5%, respectively. Multivariable analysis suggested that bleeding patients had a higher aPTT prior to the bleeding event compared with the median aPTT in non-bleeding patients (p = 0.04), and ACT and PT levels showed a trend towards a significant association (p = 0.06). Longer ECMO runs and circuit changes increased hemorrhage risk (OR = 1.004 and OR = 12.85, respectively).

Conclusion

Our study found that higher aPTT and longer ECMO duration are independent hemorrhage risk factors, underscoring the need for timely ECMO decannulation.

Keywords

child anticoagulation extracorporeal membrane oxygenation bleeding ECMO extracorporeal life support hemorrhage

Introduction

Extracorporeal membrane oxygenation (ECMO) is a life-saving therapy employed in different disease processes. While ECMO has shown improved survival amongst patients with potentially poor prognosis, there is an increased risk for complications, including thrombosis and hemorrhage.¹ The etiology stems from various factors, ranging from intrinsic and extrinsic factors, the patient’s underlying medical condition, and the physiologic response to achieving oxygenation with a mechanical device.² Thus, anticoagulation is necessary to prevent thrombosis but must be balanced with the risk of bleeding. Children are at a higher risk of bleeding while on ECMO due to multiple factors.^3–9 Several monitoring methods of anticoagulation have been employed, including a “clotting time” approach (i.e., Activated Clotting Time, ACT), Activated Partial Thromboplastin Time, aPTT), a “chromogenic” approach (i.e., Anti Activated Factor X, anti-Xa) or combination of both, along with vescoelastic assays (Thromboelastography (TEG) or Rotational thromboelastometry (ROTEM)). The Extracorporeal Life Support Organization (ELSO) has published guidelines for the management of anticoagulation on ECMO that detail the organization’s recommendations.⁷ Our objective is to correlate different anticoagulation assays with hemorrhagic events of patients who received ECMO at our center and assess whether that correlation varies with the mode of ECMO support. Since we have a combined unit and continue to care for adults with congenital heart disease, those patients were also included in our analysis. We hypothesized that higher levels of aPTT, ACT, and anti-Xa were associated with increased bleeding risk in pediatric and adult patients receiving ECMO at our center.

Materials and methods

Study design

This is a retrospective chart review of patients admitted to the Neonatal and Pediatric Intensive Care Units (NICU and PICU) at the CHRISTUS Children’s Hospital of San Antonio who required VV or VA-ECMO. Anticoagulation parameters: Activated Clotting time (ACT), anti-Xa levels, Prothrombin Time (PT), International Normalized Ratio (INR), activated Partial Thromboplastin time (aPTT), were assessed and compared against clinical outcomes. Although TEG is performed on all patients, its influence on decision-making is minimal unless results are clearly abnormal, which is why it was not included in the analysis. The primary outcome is hemorrhagic complications, defined by ELSO as a bleeding event with clinically overt bleeding recorded in the medical and nursing charts associated with either administration of 2 or more RBC units in 24 h or a drop in hemoglobin greater than 2 g/L over 24 h, and if there was a hemothorax, central nervous system or retroperitoneal bleeding, or lastly if bleeding required an intervention.⁷ A unit in pediatrics is defined as 10 ml/kg of PRBCs. If there were consecutive days with the same primary source of bleeding as preceding days, these were considered the same bleeding event. When a patient had more than one bleeding source on the same day, this was also recorded.

At our institution, we use ECMO circuits from Spectrum Medical®, including the Pedi Quadrox-iD® oxygenator for patients under 10 kg, with a ¼ inch circuit, which requires 2 units of blood and 100 ml of Fresh Frozen Plasma (FFP). For patients over 10 kg, we use the same oxygenator with a ⅜ inch circuit, requiring 3 units of blood and 150 ml of FFP. Our ECMO system incorporates the Sorin (now Liva Nova) Revolution® centrifugal pump, featuring a 94.5 g open impeller design with a 57 ml priming volume, a maximum operational speed of 3500 rpm, and a flow rate of up to 8 L/min.

Prior to cannulation, we draw CBC, PT, aPTT, fibrinogen, and antithrombin III (ATIII). ACT is measured every hour for the first 24 h post-cannulation, with subsequent adjustments based on clinical need. For the PICU, ACT is monitored every 3 h after the initial 24 h, and every 6 h as needed. In the NICU, ACT is monitored every 3 h for the first 24 h, then every 6 h as needed. Coagulation labs are drawn every 6 h, and TEG is performed every 24 h. Anti-Xa and D-dimer levels are checked every 6 h for the first 24 h, and plasma-free hemoglobin and antithrombin are checked every 24 h for 3 days.

Twenty four hours after cannulation, heparin titration is based on the anti-Xa levels, aiming for a range of 0.3-0.6 IU/ml. We administer a bolus of 10u/kg of heparin prior to giving cryoprecipitate and platelets. The heparin infusion is started at 30u/kg/hr immediately upon cannulation and titrated according to the anticoagulation protocol. Heparin doses range from 25 to 100u/kg prior to cannulation, and the heparin drip used is a 100/ml solution (25,000u/250 ml) in D5W. In the NICU, the heparin infusion is titrated using the anti-Xa heparin titration table for the first 24 h post-cannulation or after a circuit change. The NEO oxygenator is used in the NICU, while the Pedi oxygenator is used in the PICU.

We prime the ECMO circuit with RBCs, plasma, albumin (25%), heparin, NaHCO3, and CaCl2. Initially, we prime with plasmalyte, followed by blood, and add medications like NaHCO3, calcium, and heparin. For patients under 10 kg, blood priming is always used. Since the inception of the standardized ECMO program, ECMO runs, including transfusion thresholds, are protocolized and the decisions to transfuse are to a specific cutoff and will not be affected by individual decisions. The institution follows standardized transfusion targets for patients supported on ECMO to optimize hemostasis and minimize bleeding risk. Platelet counts are maintained above 100 × 10³/µL, with a lowered threshold of >80 × 10³/µL in children older than 6 years who are not requiring significant blood product support; platelets are administered at 10 mL/kg. Red blood cell transfusions are guided by a hematocrit goal of >30%, with higher targets (>35%) applied in neonates, postoperative cardiac patients, and those with cyanotic heart disease; RBCs are dosed at 10 mL/kg. Cryoprecipitate is used to maintain fibrinogen levels above 200 mg/dL, with an adjusted goal of >150 mg/dL in patients with minimal blood product utilization, administered as one unit per 10 kg body weight. Antithrombin III replacement is generally considered when levels fall below 50%, although Thrombate use is reserved for cases with high heparin requirements or suspected heparin resistance, and routine use of FFP for low antithrombin levels is not standard. While the NICU applies a similar 50% cutoff, they tend to be more proactive in administering Thrombate.

The study was approved by the Institutional Review Board for Human Subject Research for Baylor College of Medicine and Affiliated Hospitals (BCM IRB-H-51,520).

Study population

Patients admitted to the PICU and NICU at CHRISTUS Children’s of San Antonio from 2014 to 2023 who were with VV or VA-ECMO. Our PICU is a combined unit that also cares for adult congenital heart disease patients, some of whom require ECMO treatment. The patient population is heterogeneous, with wide age ranges and different pathologies.

Inclusion criteria

(1) PICU and NICU admissions at CHRISTUS Children’s of San Antonio.

(2) VV or VA-ECMO initiation.

(3) Anticoagulation using heparin.

Exclusion criteria

(1) Patients with a previously identified history of bleeding disorder.

(2) Patients with bleeding episodes that happened before the initiation of ECMO and discovered within 24 h of ECMO initiation.

(3) Patients receiving direct thrombin antagonists (DTA) for anticoagulation. These patients are monitored using aPTT only, and the use of DTAs is infrequent at our institution.

Statistical analysis

Descriptive data are reported as frequencies and proportions for categorical variables and median (interquartile range, IQR) for continuous variables. Differences between groups were determined by the Wilcoxon rank sum test for the continuous variables and chi-square or Fisher’s exact tests for the categorical variables. The cumulative incidence of hemorrhagic events was depicted by the Kaplan-Meier (KM) curves. Levels of select coagulation parameters were presented by the box plots. Logistic regression was used to determine coagulation factors associated with hemorrhagic events and mortality. Linear regression was used to determine the coagulation factors associated with the length of stay (LOS). Because of the heterogeneity of the cohort and the possibility of confounders, two multivariable models were constructed for each regression analysis, each with a different anticoagulant measurement used for those who experienced a hemorrhagic event. The median anticoagulant parameter was used in one model, while the other used the anticoagulant measurement taken within 6 h of the hemorrhagic event. Variable selections for the multivariable models were conducted using the least absolute shrinkage and selection operator (Lasso) method with the cross-validation (CV) selection option or clinical importance.^10,11 Briefly, the Lasso program was run with all variables included in the univariable analysis and suggested good initial models, which included the variables with a high probability of being risk factors. Covariates were also reviewed by clinicians for clinical relevance. The best model was selected based on the smallest Bayesian Information Criterion (BIC). All analyses were performed on Stata version 18.5 (StataCorp LLC, College Station, TX, USA).¹⁰ A p-value of <0.05 was considered statistically significant.

Results

Our study included 76 patients admitted to the PICU and NICU at the CHRISTUS Children’s Hospital of San Antonio between 2014 and 2023. In total, 32/76 (42.1%) patients experienced a hemorrhagic event. Among those who experienced a hemorrhagic event, 15 patients (46.9%) survived. A larger proportion of survivors, 30/44 (68.2%), was observed among those who did not experience a hemorrhagic event (Figure 1).

Figure 1.

Flowchart diagram of study selection.

The median age of the patients was 1.1 months (IQR 0.0-46.8). Most patients, 59 (77.6%), received VA-ECMO. The median cannulation hours were 143.5 (IQR 84.5-294.0). Patients with hemorrhagic events had a higher rate of circuit changes compared to patients who did not have hemorrhagic events (18.8% vs 2.3%, p = .01) (Table 1).

Table 1.

Patient characteristics, by bleeding event.

	Total	No bleeding	Bleeding	p-value
	(N = 76)	(n = 44)	(n = 32)	p-value
Age in months, median (IQR)	1.1 (0.0, 46.8)	0.7 (0.0, 19.9)	2.6 (0.0, 60.2)	0.31
Age in days, median (IQR)	33.0 (1.0, 1423.5)	22.0 (1.0, 603.5)	80.0 (1.5, 1830.0)	0.31
Cannulation hours, median (IQR)	143.5 (84.5, 294.0)	140.5 (91.5, 241.5)	156.5 (80.5, 423.0)	0.52
Circuit change?				0.01
No	69 (90.8)	43 (97.7)	26 (81.2)
Yes	7 (9.2)	1 (2.3)	6 (18.8)
Child development categories				0.42
Pediatrics	35 (46.1)	22 (50.0)	13 (40.6)
Neonates	41 (53.9)	22 (50.0)	19 (59.4)
VA vs. VV ECMO				0.64
VA	59 (77.6)	35 (79.5)	24 (75.0)
VV	17 (22.4)	9 (20.5)	8 (25.0)
ACT median, median (IQR)	162.1 (152.5, 175.0)	162.1 (152.5, 176.0)	161.2 (150.0, 170.0)	0.51
Anti-XA median, median (IQR)	0.4 (0.3, 0.5)	0.4 (0.3, 0.5)	0.3 (0.3, 0.5)	0.23
INR median, median (IQR)	1.2 (1.1, 1.3)	1.2 (1.1, 1.3)	1.2 (1.1, 1.4)	0.40
PT median, median (IQR)	15.1 (14.1, 16.6)	15.2 (14.1, 16.4)	15.1 (14.1, 17.4)	0.84
aPTT median, median (IQR)	96.4 (80.5, 119.5)	93.2 (78.8, 109.0)	99.0 (84.9, 137.1)	0.23
ACT median vs ACT prior bleed, median (IQR)	164.5 (154.4, 177.5)	162.1 (152.5, 176.0)	167.0 (158.0, 184.5)	0.07
Anti-XA median vs Anti-XA prior bleed, median (IQR)	0.4 (0.3, 0.5)	0.4 (0.3, 0.5)	0.4 (0.2, 0.7)	0.74
Anti-Xa median vs anti-Xa prior bleed				0.20
<0.25	11 (14.5)	4 (9.1)	7 (21.9)
0.25-0.5	53 (69.7)	34 (77.3)	19 (59.4)
>0.5	12 (15.8)	6 (13.6)	6 (18.8)
INR median vs INR prior bleed, median (IQR)	1.2 (1.0, 1.4)	1.2 (1.1, 1.3)	1.2 (1.0, 1.5)	0.42
INR median vs INR prior bleed				0.21
<1.2	43 (57.3)	24 (54.5)	19 (61.3)
1.2-1.5	24 (32.0)	17 (38.6)	7 (22.6)
>1.5	8 (10.7)	3 (6.8)	5 (16.1)
PT median vs PT prior bleed, median (IQR)	15.6 (14.0, 17.0)	15.2 (14.1, 16.5)	16.2 (13.9, 18.8)	0.29
aPTT median vs aPTT prior bleed, median (IQR)	99.6 (80.5, 141.0)	93.2 (78.8, 108.9)	124.9 (84.8, 250.0)	0.02
Total amount of cryo transfused, median (IQR)	0.0 (0.0, 2.7)	0.0 (0.0, 2.4)	0.0 (0.0, 2.7)	0.27
Total amount of FFP transfusion, median (IQR)	21.9 (0.0, 63.6)	20.9 (0.0, 63.6)	21.9 (0.0, 70.0)	0.73
Total amount of platelet transfusion, median (IQR)	22.5 (0.0, 74.5)	12.9 (0.0, 109.6)	28.6 (8.5, 64.5)	0.60
Total amount of PRBC transfusion, median (IQR)	39.2 (9.5, 132.4)	44.2 (8.8, 122.2)	39.2 (15.9, 149.8)	0.76
Total blood product, median (IQR)	151.1 (22.8, 318.9)	157.9 (13.7, 352.0)	149.1 (43.8, 294.1)	0.67
LOS^a (in patients who are alive), median (IQR)	47.0 (26.0, 76.0)	49.0 (28.0, 83.0)	43.0 (25.0, 61.0)	0.52
Expired?				0.06
No	45 (59.2)	30 (68.2)	15 (46.9)
Yes	31 (40.8)	14 (31.8)	17 (53.1)

^aLOS information is available for n = 45 survivors.

Figure 2 presents the time to events estimated by the Kaplan-Meier statistic. The estimated cumulative incidence of the hemorrhagic event at 10-, 20-, and 30-days post-cannulation was 36.9% (26.2%-50.3%), 58.8% (41.9%-76.5%), and 72.5% (47.2%-92.7%), respectively (Figure 2, Table S1). Sub-analysis indicated that patients with circuit change had 85.7% bleeding compared with 37.7% in those who did not have circuit change (Table S2). No difference in the bleeding proportion was found between VV versus VA ECMO groups (Table S3) nor between neonates versus pediatrics (Table S4).

Figure 2.

Kaplan-Meir plot of hemorrhagic events.

We compared the median values of anticoagulant parameters, starting 24 h after ECMO cannulation, between patients who experienced a hemorrhagic event and those who did not. We found no significant differences between the two groups (Table 1). The medians excluded the values collected in the first 24 h after cannulation because a heparin bolus is given right before cannulation; thus, anticoagulation parameters are expected to be elevated during the first 24 h after cannulation.

We subsequently compared the median anticoagulant parameters measured from patients who did not experience a hemorrhagic event to the median anticoagulant parameters measured prior to the hemorrhagic event (within 6 h of the event) for those who experienced a bleeding event. Our results indicated a difference in median aPTT and median aPTT prior to the bleeding event (p = 0.02) (Figure 3). Patients having hemorrhagic events appeared to have a higher median ACT level compared with the median ACT prior to the bleeding event in the group without hemorrhagic events. However, the difference did not reach statistical significance (p = 0.07) (Figure S1). The remaining comparisons showed no significant differences (Table 1).

Figure 3.

Boxplot of aPTT by bleeding event; (A) median aPTT from non-bleed patients versus median aPTT for bleed patients; (B) median aPTT from non-bleed patients versus aPTT-prior-bleeding-event from bleed patients.

Logistic regression analysis of hemorrhagic event

The results of the two logistic regression models for hemorrhagic events indicate that a unit increase in cannulation hours increases the odds ratio up to 1% (OR = 1.004 [1.001-1.01], p = 0.02; (OR = 1.01 [1.001-1.01], p = 0.004) (Table 2). Similarly, circuit change increased the likelihood of experiencing the event (OR = 12.85 [1.31-126.56], p = 0.03; OR = 15.82 [1.37-182.16], p = 0.03). aPTT median versus aPTT prior to the bleeding event was also associated with bleeding (p = 0.04), while ACT and PT levels showed a trend towards a significant association (p = 0.06) (Table 2). Model 1 did not attribute any significant effects to anticoagulant factors; however, model 2 associated aPTT with an estimated increase in the odds ratio of 1% (OR = 1.01 [1.002-1.03], p = 0.03) (Table 2).

Table 2.

Multivariable logistic regression for bleeding event.

	Model 1		Model 2
	OR (95% CI)	p-values	OR (95% CI)	p-values
Age in months, median (IQR)	1.00 (0.998, 1.01)	0.24	1.00 (0.99, 1.01)	0.69
Cannulation hours, median (IQR)	1.004 (1.001, 1.01)	0.02	1.01 (1.001, 1.01)	0.004
Circuit change	12.85 (1.31, 126.56)	0.03	15.82 (1.37, 182.16)	0.03
ACT median	0.97 (0.94, 1.002)	0.06
PT median	1.23 (0.99, 1.54)	0.06
aPTT median	1.01 (0.997, 1.03)	0.13
ACT median vs ACT prior to bleed			1.00 (0.99, 1.02)	0.57
PT median vs PT prior to bleed			1.14 (0.89, 1.45)	0.30
aPTT median vs aPTT prior to bleed			1.02 (1.001, 1.03)	0.04
Anti-Xa median vs anti-Xa prior to bleed			0.77 (0.11, 5.68)	0.80
	AUC = 0.80		AUC = 0.85

OR, odds ratio; CI, confidence interval; AUC, area under the curve.

Model 1: coagulation parameters are their medians in both groups, with and without bleeding event.

Model 2: coagulation parameters are medians in non-bleeding group and values prior bleeding event in patients who had bleeding event.

Logistic regression analysis for mortality

Both multivariable logistic models for mortality suggested a 1% increase in odds ratio due to cannulation hour (OR = 1.01 [1.003-1.02], p = 0.002; OR = 1.01 [1.004-1.02], p = 0.001) (Table 3). In contrast, both models attributed a decrease in the likelihood of mortality to VV-ECMO (OR = 0.09 [0.01-0.91], p = 0.04; OR = 0.11 [0.01-0.93], p = 0.04) (Table 3). Although model 2 did not attribute a significant effect to any anticoagulant parameters, model 1 associated PT with a 57% increase in odds ratio (OR = 1.57 [1.13-2.18], p = 0.01) (Table 3).

Table 3.

Multivariable logistic regression for mortality.

	Model 1		Model 2
	OR (95% CI)	p-values	OR (95% CI)	p-values
Age in months, median (IQR)	1.01 (0.999, 1.01)	0.09	1.00 (0.997, 1.01)	0.30
Bleed	0.97 (0.23, 4.13)	0.970	0.60 (0.14, 2.56)	0.49
Cannulation hours, median (IQR)	1.01 (1.003, 1.02)	0.002	1.01 (1.004, 1.02)	0.001
VA v VV ECMO	0.09 (0.01, 0.91)	0.04	0.11 (0.01, 0.93)	0.04
Circuit change	14.22 (0.82, 247.73)	0.07	9.09 (0.64, 128.65)	0.10
Total amount of FFP transfusion	1.00 (0.98, 1.01)	0.41	0.99 (0.98, 1.004)	0.25
ACT median	1.01 (0.97, 1.06)	0.59
PT median	1.57 (1.13, 2.18)	0.01
aPTT median	1.00 (0.98, 1.02)	0.81
ACT median vs ACT prior to bleed			1.00 (0.99, 1.02)	0.51
PT median vs PT prior to bleed			1.30 (0.99, 1.72)	0.07
aPTT median vs aPTT prior to bleed			1.01 (0.99, 1.02)	0.47
Anti-Xa median vs anti-Xa prior to bleed			0.50 (0.05, 4.83)	0.55
	AUC = 0.88		AUC = 0.86

OR, odds ratio; CI, confidence interval; AUC, area under the curve.

Model 1: coagulation parameters are their medians in both groups, with and without bleeding event.

Model 2: coagulation parameters are medians in non-bleeding group and values prior bleeding event in patients who had bleeding event.

Linear regression analysis for LOS

The linear regression models for LOS indicated a positive association with cannulation hours (0.22 [0.07-0.37], p = 0.004; 0.16 [0.01-0.32], p = 0.04) (Table 4). Neither model indicated a relationship between LOS and any of the anticoagulant parameters.

Table 4.

Multivariable linear regression for length of stay (includes only those that did not expire, n = 45).

	Model 1		Model 2
	Coef. (95% CI)	p-values	Coef. (95% CI)	p-values
Age in months, median (IQR)	−0.15 (−0.37, 0.07)	0.17	−0.12 (−0.34, 0.11)	0.30
Bleed	−2.15 (−39.32, 35.02)	0.91	20.11 (−21.65, 61.87)	0.34
Cannulation hours, median (IQR)	0.22 (0.07, 0.37)	0.004	0.16 (0.01, 0.32)	0.04
VA v VV ECMO	−21.58 (−57.00, 13.84)	0.23	−36.83 (−73.41, 0.24)	0.049
PT median	6.16 (−3.84, 16.16)	0.22
aPTT median	0.37 (−0.13, 0.86)	0.15
PT median vs PT prior to bleed			−0.21 (−9.70, 9.28)	0.96
aPTT median vs aPTT prior to bleed			−0.14 (−0.53, 0.26)	0.49
Anti-Xa median vs anti-Xa prior to bleed			9.98 (−41.45, 61.40)	0.70

CI, confidence interval.

Model 1: coagulation parameters are their medians in both groups, with and without bleeding event.

Model 2: coagulation parameters are medians in non-bleeding group and values prior bleeding event in patients who had bleeding event.

Linear regression analysis for the length of ECMO

Both linear regression models for ECMO hours found a positive association with age (0.64 [0.18-1.11], p = 0.01; 0.51 [0.04-0.98], p = 0.03) (Table 5).

Table 5.

Univariable and multivariable linear regression for cannulation hours (i.e., ECMO run hours).

	Model 1		Model 2
	Coef (95% CI)	p-value	Coef (95% CI)	p-value
Age in months, median (IQR)	0.64 (0.18, 1.11)	0.01	0.51 (0.04, 0.98)	0.03
Anti-Xa median	162.11 (−174.92, 499.15)	0.34
PT median	−6.70 (−22.08, 8.68)	0.39
aPTT median	−1.63 (−2.99, −0.28)	0.02
Anti-Xa median vs anti-Xa prior to bleed			165.56 (15.38, 315.74)	0.03
PT median vs PT prior to bleed			−6.92 (−20.81, 6.97)	0.32
aPTT median vs aPTT prior to bleed			−1.45 (−2.35, −0.54)	0.002

CI, confidence interval.

Model 1: coagulation parameters are their medians in both groups, with and without bleeding event.

Model 2: coagulation parameters are medians in non-bleeding group and values prior bleeding event in patients who had bleeding event.

Among the anticoagulant parameters, both models suggested a significant negative association with aPTT (−1.63 [−2.99, −0.28], p = 0.02; −1.45 [−2.35 to −0.54], p = 0.002). In addition, model 2 indicated a significant effect from median anti-Xa and median anti-Xa prior to bleeding (165.56 [15.38-315.74], p = 0.03) (Table 5).

Discussion

Our study demonstrated that, for patients who were placed on VV or VA-ECMO, a higher aPTT was associated with a significantly higher risk of bleeding. A higher ACT and PT were also associated with a higher risk of bleeding, but the difference was not statistically significant. More importantly, the study shows that the longer the patient was on ECMO, the higher the probability of having a hemorrhagic event. Mortality was higher in the bleeding group, but no statistical significance was reached. Interestingly, circuit change was associated with bleeding events. This could be due to our practice of giving a heparin bolus after a circuit change. This suggests that hemorrhagic events are multifactorial, and future studies should consider other variables.

Since the inception of extracorporeal life support, anticoagulation has been the cornerstone of managing the inevitable thrombotic risk as part of using this modality of treatment. After successfully performing cardiopulmonary bypass (CPB) using a membrane oxygenator, extracorporeal support was maintained for 4 days in dogs in the 1960s.^12–14 As in CPB, heparin was used to prevent clotting in the pump and circuit. The bleeding was significant, and the heparin dose was titrated in stages while monitoring using only ACT.¹⁵ The first successful clinical cases of ECMO were reported in the early 1970s.^4,16

The pathophysiology of clotting during ECMO is considerably complex. Despite significant advances, our current understanding of the biological mechanisms underlying clotting remains limited. This gap in knowledge likely contributes to the ongoing challenges in developing effective anticoagulation strategies. Once the blood contacts the plastic tubing of the ECMO circuit, a cascade of systemic and cellular events will trigger an inflammatory response. These include several humoral and cellular systems, including contact, intrinsic, and extrinsic coagulation, complement systems, endothelial cells, leukocytes, platelets, and cytokines.² In the absence of anticoagulation, this will inevitably lead to thrombin deposition and the formation of clots. Clotting is not confined to the circuit; it can also occur in the patient’s blood vessels and cause significant morbidity and possibly mortality.

ELSO acknowledges that a balance between anticoagulation and pro-coagulant function is necessary to avoid issues with inappropriate hemorrhage and thrombosis.⁷ In a multicenter study in 8 children’s hospitals, the authors reported bleeding events in 70.2% of children receiving ECMO, including intracranial hemorrhage in 16.0%. The bleeding events were independently associated with a higher daily risk of mortality.¹⁷ A systematic review reported a third of the patients had a bleeding event of any severity.¹⁸ Sy et al. reported the incidence of major bleeding to range between 13% and 50% in a meta-analysis performed on observational studies that included 1496 patients supported with VA-ECMO, and it was correlated to the strategy used to monitor heparin.¹⁹ Bleeding complications were reported at 27.6% in an observational study of the ELSO registry.²⁰ Two-thirds of those who experienced bleeding did not survive to hospital discharge, compared to less than half of those who did not experience bleeding.²⁰ We report 42.1% hemorrhagic events associated with ECMO in our cohort. Less than half of those patients survived to hospital discharge. Other studies have reported that bleeding has been associated with worse outcomes.^17,20,21

ELSO Guidelines were published in 2022 to assist physicians in managing anticoagulation on ECMO.⁷ Ideally, monitoring would be conducted in vivo, which captures a comprehensive instantaneous view of a patient’s coagulation. However, out of necessity, the assays commonly employed are in vitro and typically assess only part of the coagulation cascade. This, in turn, leads to an imperfect assessment of the patient’s true coagulation potential. Additionally, the recommendation from ELSO for optimal anticoagulation states that there is no universally accepted best practice, and each facility must find what serves its patients best.⁷ This is supported by several other studies that cite the inter-facility differences in anticoagulation goals, titration, and dosing of anticoagulants.^22–28

The challenge is higher when caring for neonates undergoing ECMO. There is a poor correlation among laboratory assays used in anticoagulation monitoring. This has led to center-specific algorithms for coagulation monitoring labs and the frequency of obtaining them. To help guide decision-making, it is important to evaluate the influential factors that will affect coagulation.^24,29–31

There are two ways in which anticoagulation can be assessed. The first is measuring the time it takes the blood to form a clot, i.e., clotting time. This is based on either standard laboratory testing: aPTT or the whole blood point-of-care ACT. The second measures heparin concentration by chromogenic approach, based on anti-Xa activity.³² Many of the standard tests used to assess anticoagulation are site-dependent. This could be attributed to the variability among labs when performing those tests. For example, anti-Xa assay results vary depending on whether the performing lab adds antithrombin to the sample.³³ This will lead to site-dependent variability and reproducibility. Thus, the standard of care anti-Xa levels at one site cannot be reliably translated as a gold standard to other sites.³³

Similarly, aPTT is another commonly used test to assess anticoagulation, and based on what reagents are used in the performing laboratory, the reported time can be affected.³² This makes inter-facility comparisons challenging.^23,26 Similar to our results, Sklar et al. suggested that a lower aPTT may aid in reducing the risk of life-threatening hemorrhagic events on VV-ECMO. This systemic analysis found that, while there was an increased risk of developing thrombotic events, this risk was mainly related to the circuit. They concluded that an aPTT under 60 s should be targeted in this population of ECMO patients.²³

Regarding ACT, the activator used will affect its reproducibility and clot detection.^34,35 Also, variability was observed within the same facility, as the technique is dependent on the operator.³⁵ Finally, variables such as temperature, hemodilution, hematocrit, fibrinogen level, and platelet count and function will affect the results as the assay is performed on whole blood.^34,35

Vescoelastography (TEG® and ROTEM®) has been used to assess ECMO anticoagulation in recent decades. Those assays provide a detailed analysis of the blood’s ability to form a clot. They also measure the time to clotting initiation and clot strength. Their use is limited by the time and cost consumed to perform the assay. Not enough studies correlate those assays with outcomes or the risk of bleeding on ECMO. Those assays can be beneficial in detecting underlying coagulopathy after adding a heparinase.^36–40 This subset of patients could have other abnormal parameters, such as higher INR and a prolonged aPTT. Because these parameters could be altered due to various reasons, clinical management decisions based on TEG results were primarily applied to patients with suspected or confirmed underlying bleeding disorders at our institution, who represented a small subset of the overall cohort.

Specific goals of anticoagulation will vary depending on the perceived risk of bleeding. This leads to complexity in comparing one patient to another in terms of what anticoagulation goals should be accepted.^7,24,25 In a recent study, COVID-19 patients receiving ECMO were found to have a higher frequency of coagulation disorders.²¹

Moreover, children and neonates are at higher risk of bleeding due to the different concentrations of coagulation factors and their response to heparin.^24,29–31 As such, no one test can reliably coordinate optimal anticoagulation while on ECMO.^7,24,26,27

Conclusion

Our study identified several factors associated with hemorrhagic complications, mortality, and ECMO duration. Elevated aPTT and prolonged ECMO duration were independently associated with an increased risk of bleeding events. Cannulation time and circuit change were also found to increase bleeding risk, while PT showed a trend toward significance. For mortality, prolonged cannulation duration was associated with higher odds of death, whereas VV-ECMO configuration was linked to improved survival. Although coagulation parameters had limited predictive value for mortality, PT demonstrated a modest association. Additionally, older age and higher anti-Xa levels correlated with longer ECMO duration.

These findings highlight the multifactorial nature of coagulation and bleeding risks in pediatric ECMO. Tailoring anticoagulation strategies to patient-specific and pathology-specific factors remains essential. Future research should focus on differentiating ECMO phenotypes and addressing thromboinflammatory pathways rather than relying solely on traditional coagulation assays.

Study limitations

This is a retrospective study conducted by a single institution on the available database. It should be interpreted with some limitations in mind. Observational studies play a significant role in health research, particularly when evidence from randomized controlled experiments is not available or feasible. However, retrospective analyses are prone to recall, or misclassification bias and chronological relationships are often difficult to assess. Moreover, some risk factors may be present but cannot be measured, predisposing such studies to misinterpretation. A limitation of this study is the complexity of the pathophysiology of clotting during ECMO, which is not fully understood. The underlying pathologies were not consistently documented during the data collection and were thus not included. Variability in underlying diagnoses could influence coagulopathy and bleeding risk. Future investigation in prospective studies with more granular diagnostic data collection. Despite advances in technology and clinical practice, the limited knowledge of the biological mechanisms of clotting remains a challenge, complicating the development of effective anticoagulation strategies.

Supplemental material

Supplemental material - Correlation of anticoagulation assays with bleeding during extracorporeal membrane oxygenation at a tertiary pediatric hospital

Supplemental Material for Correlation of anticoagulation assays with bleeding during extracorporeal membrane oxygenation at a tertiary pediatric hospital by Mohammed Salameh, Jazmin Delgado, Ryan Killebrew, Cody Henderson, Elumalai Appachi in Perfusion

Footnotes

Acknowledgments

The authors acknowledge the essential contributions of the Biostatistics team at Baylor College of Medicine, namely Duc T Nguyen, MD, PhD, Kenneth Jhune Nobleza, and Ankona Banerjee.

ORCID iD

Mohammed Salameh

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental material

Supplemental material for this article is available online.

Appendix

References

Tonna

Boonstra

MacLaren

, et al. Extracorporeal life support organization registry international report 2022: 100,000 survivors. ASAIO J 2024; 70: 131–143.

Millar

Fanning

McDonald

, et al. The inflammatory response to extracorporeal membrane oxygenation (ECMO): a review of the pathophysiology. Crit Care 2016; 20: 387.

Barbaro

Paden

Guner

, et al. Pediatric extracorporeal life support organization registry international report 2016. ASAIO J 2017; 63: 456–463.

Bartlett

Roloff

Cornell

, et al. Extracorporeal circulation in neonatal respiratory failure: a prospective randomized study. Pediatrics 1985; 76: 479–487.

Jenks

Landry

Garrison

, et al. Pediatric extracorporeal membrane oxygenation anticoagulation protocol associated with a decrease in complications. ASAIO J 2022; 68: 275–280.

Jenks

Raman

Dalton

. Pediatric extracorporeal membrane oxygenation. Crit Care Clin 2017; 33: 825–841.

Mcmichael

BVA

Ryerson

Ratano

, et al. 2021 ELSO adult and pediatric anticoagulation guidelines. ASAIO J 2022; 68: 303–310.

Paden

Conrad

Rycus

, et al. Extracorporeal life support organization registry report 2012. ASAIO J 2013; 59: 202–210.

Rabinowitz

Danzo

Anderson

, et al. Anticoagulation-free pediatric extracorporeal membrane oxygenation: single-center retrospective study. Pediatr Crit Care Med 2023; 24: 499–509.

10.

StataCorp . Stata 18 lasso reference manual. Stata Press.: StataCorp LLC, 2023.

11.

Hastie

Wainwright

RTM

. Statistical learning with sparsity:the lasso and generalizations. 1 ed. CRC Press, 2015.

12.

Bartlett

Drinker

Burns

, et al. The toroidal membrane oxygenator: design, performance, and prolonged bypass testing of a clinical model. Trans Am Soc Artif Intern Organs 1972; 18: 369–374.

13.

Bartlett

Isherwood

Moss

, et al. A toroidal flow membrane oxygenator: four day partial bypass in dogs. Surg Forum 1969; 20: 152–153.

14.

Bartlett

Kittredge

Noyes

Jr. , et al. Development of a membrane oxygenator: overcoming blood diffusiolimitation. J Thorac Cardiovasc Surg 1969; 58: 795–800.

15.

Lenahan

Phillips

. Some variables which influence the activated partial thromboplastin time assay. Clin Chem 1966; 12: 269–273.

16.

Bartlett

Gazzaniga

Toomasian

, et al. Extracorporeal membrane oxygenation (ECMO) in neonatal respiratory failure. 100 cases. Ann Surg 1986; 204: 236–245.

17.

Dalton

Reeder

Garcia-Filion

, et al. Factors associated with bleeding and thrombosis in children receiving extracorporeal membrane oxygenation. Am J Respir Crit Care Med 2017; 196: 762–771.

18.

Olson

Murphree

Zonies

, et al. Thrombosis and bleeding in extracorporeal membrane oxygenation (ECMO) without anticoagulation: a systematic review. ASAIO J 2021; 67: 290–296.

19.

Sklar

Lequier

, et al. Anticoagulation practices and the prevalence of major bleeding, thromboembolic events, and mortality in venoarterial extracorporeal membrane oxygenation: a systematic review and meta-analysis. J Crit Care 2017; 39: 87–96.

20.

Willers

Swol

Buscher

, et al. Longitudinal trends in bleeding complications on extracorporeal life support over the past two decades-extracorporeal life support organization registry analysis. Crit Care Med 2022; 50: e569–e580.

21.

Feth

Weaver

Fanning

, et al. Hemorrhage and thrombosis in COVID-19-patients supported with extracorporeal membrane oxygenation: an international study based on the COVID-19 critical care consortium. J Intensive Care 2024; 12: 18.

22.

Anticoagulation task force.

23.

Sklar

Lequier

, et al. Anticoagulation practices during venovenous extracorporeal membrane oxygenation for respiratory failure. A systematic review. Ann Am Thorac Soc 2016; 13: 2242–2250.

24.

Kamdar

Rintoul

Raffini

. Anticoagulation in neonatal ECMO. Semin Perinatol 2018; 42: 122–128.

25.

Koster

Ljajikj

Faraoni

. Traditional and non-traditional anticoagulation management during extracorporeal membrane oxygenation. Ann Cardiothorac Surg 2019; 8: 129–136.

26.

Chlebowski

Baltagi

Carlson

, et al. Clinical controversies in anticoagulation monitoring and antithrombin supplementation for ECMO. Crit Care 2020; 24: 19.

27.

Descamps

Moussa

Besnier

, et al. Anti-Xa activity and hemorrhagic events under extracorporeal membrane oxygenation (ECMO): a multicenter cohort study. Crit Care 2021; 25: 127.

28.

Richard

Argaud

Blet

, et al. Extracorporeal life support for patients with acute respiratory distress syndrome: report of a consensus conference. Ann Intensive Care 2014; 4: 15.

29.

Stocker

Horton

. Anticoagulation strategies and difficulties in neonatal and paediatric extracorporeal membrane oxygenation (ECMO). Perfusion 2016; 31: 95–102.

30.

Miller

Bailey

Mancuso

, et al. Functional maturity of the coagulation system in children: an evaluation using thrombelastography. Anesth Analg 1997; 84: 745–748.

31.

Oliver

. Anticoagulation and coagulation management for ECMO. Semin CardioThorac Vasc Anesth 2009; 13: 154–175.

32.

Bates

Weitz

. Coagulation assays. Circulation 2005; 112: e53–e60.

33.

Holbrook

Schulman

Witt

, et al. Evidence-based management of anticoagulant therapy: Antithrombotic therapy and prevention of thrombosis, 9th ed: American college of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 2012; 141: e152S–e184S.

34.

Despotis

Gravlee

Filos

, et al. Anticoagulation monitoring during cardiac surgery: a review of current and emerging techniques. Anesthesiology 1999; 91: 1122–1151.

35.

Ojito

Hannan

Burgos

, et al. Comparison of point-of-care activated clotting time systems utilized in a single pediatric institution. J Extra Corpor Technol 2012; 44: 15–20.

36.

Harr

Moore

Chin

, et al. Viscoelastic hemostatic fibrinogen assays detect fibrinolysis early. Eur J Trauma Emerg Surg 2015; 41: 49–56.

37.

Sankarankutty

Nascimento

Teodoro da Luz

, et al.

TEG(R) and ROTEM(R) in trauma: similar test but different results?

World J Emerg Surg 2012; 7(Suppl 1): S3.

38.

Tomori

Hupalo

Teranishi

, et al. Evaluation of coagulation stages of hemorrhaged swine: comparison of thromboelastography and rotational elastometry. Blood Coagul Fibrinolysis 2010; 21: 20–27.

39.

Ziegler

Voelckel

Zipperle

, et al. Comparison between the new fully automated viscoelastic coagulation analysers TEG 6s and ROTEM sigma in trauma patients: a prospective observational study. Eur J Anaesthesiol 2019; 36: 834–842.

40.

Stammers

Willett

Fristoe

, et al. Coagulation monitoring during extracorporeal membrane oxygenation: the role of thrombelastography. J Extra Corpor Technol 1995; 27: 137–145.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.72 MB