Secondary Logo

Journal Logo

Pediatric Circulatory Support

Impact of Anticoagulation and Circuit Technology on Complications During Extracorporeal Membrane Oxygenation

Niebler, Robert A.*; Parker, Hinah*; Hoffman, George M.*,†

Author Information
doi: 10.1097/MAT.0000000000000811
  • Free


Patients on extracorporeal membrane oxygenation (ECMO) support require systemic anticoagulation to prevent thrombus formation and allow for uninterrupted blood flow through the circuit. The need for systemic anticoagulation increases the risk of hemorrhagic complications.1–4 There is no laboratory test that has been found to have predictive value concerning thrombotic and hemorrhagic complications, and thus the approach to anticoagulation remains multifactorial and varies across institutions.5–7

Traditional ECMO anticoagulation protocols have focused on using a goal activated clotting time (ACT) to titrate heparin dosing in an effort to keep the balance between excessive pro- and anticoagulation.1 In addition to its convenience as a bedside test, it requires a smaller blood volume and provides a more rapid response than other testing.8 However, multiple studies have found a lack of reliability when looking at ACT as an indicator of heparin level or effect.8,9 Despite these limitations, ACT has been, and continues to be, the most commonly used test of anticoagulation for children on ECMO.10 Inconsistencies arise when looking at the relationship of ACT to heparin dosing, which is affected by multiple factors such as thrombocytopenia, low fibrinogen, hypothermia, and hemodilution which all prolong ACT independent of heparin effects.11 Discrepancy between measurement systems and proportional reliance on ACT for heparin dosing, adds additional inconsistency when comparing data across various institutions.5,12

Extracorporeal membrane oxygenation circuit technology has evolved over time. In particular, pump and oxygenator technology have changed significantly. Many programs have transitioned from the use of roller pumps to centrifugal pumps despite evidence of increased rates of complications including hemolysis and kidney injury.13,14 Circuit coating technology has been adapted to improve the biocompatibility of the circuit, but the effectiveness of these coatings is questionable.15 New oxygenator technology has resulted in lower priming volumes and reduced pressure drop,16 but no published report has demonstrated improved clinical outcomes.

In June 2011, the ECMO heparin management protocol at the Children’s Hospital of Wisconsin was changed to target antifactor Xa (anti-Xa) levels. Anit-Xa levels require a greater blood volume, cannot be done bedside, and take longer to result.8 Despite these drawbacks, anit-Xa levels have been shown to have a better correlation with heparin dosing.8 At the same time the circuit technology changed from a mixture of roller pumps with a first-generation centrifugal pump and a heparin bonded circuit to a newer generation magnetically levitated centrifugal pump with an albumin/heparin bonded circuit and a newer generation oxygenator. The objective of this study is to evaluate the effect these changes had on subsequent hemorrhagic and thrombotic complications related to ECMO support.

Materials and Methods

Study Design and Approval

An institutional review board-approved retrospective chart review was performed on all patients at Children’s Hospital of Wisconsin receiving ECMO support from January 2006 through March 2016. The 3 months immediately preceding and following the changes in June 2011 (March through September 2011) were excluded to allow for complete transition from the old to new protocol and technology. Treatment cohorts were defined by the primary heparin management strategy (ACT target before protocol change; anti-Xa target after protocol change).

Extracorporeal Membrane Oxygenation Circuit Composition and Activated Clotting Time Measurement Device

The ECMO circuit composition changed over the time of the study period. In the ACT target group, 2 types of pumps were used. A roller pump (Stockert S3 Roller Pump; Stockert Instruments, Munich, Germany) with a bladder reservoir (Better Bladder; Circulatory Technology Inc, Oyster Bay, NY, USA) was used as the standard circuit. An emergency response circuit consisting of a centrifugal pump (Biomedicus Bio-Pump BP-80; Medtronic Inc., Minneapolis, MN, USA) without a reservoir was used for rapid deployment in select patients. The circuits utilized in this period were heparin bonded (Carmeda AB; W.L. Gore & Associates, Upplands Vasby, Sweeden). A heparin-coated hollow fiber membrane oxygenator (Minimax Plus Oxygenator; Medtronic Inc. or Quadrox D Oxygenator; MAQUET GmbH & Co. KG, Wayne, NJ, USA) was used for all circuits. In the anti-Xa group, the circuit consisted solely of a centrifugal pump (ROTAFLOW Centrifugal Pump; MAQUET GmbH & Co. KG) without a reservoir. The circuit coating changed to an albumin- and heparin-bonded coating (BIOLINE Coating; MAQUET GmbH & Co. KG). The oxygenator was changed as well to a heparin-coated hollow fiber membrane oxygenator (Quadrox iD Adult or iD Pediatirc for patients less than 10 kg; MAQUET GmbH & Co. KG).

Activated clotting time measurements were performed with the Hemochron Jr and Hemochron Signature Elite (International Technidyne Corporation, Edison, NJ, USA).

Anticoagulation and Transfusion Protocols

The ACT target protocol consisted of a nurse-driven titration of heparin infusion to meet physician ordered ACT goals. Activated clotting time measurements were drawn at least hourly and more frequently if changes were made within the hour. The ACT goals were driven by patient risk stratification as determined by the treating clinician. An ACT of 180–200 seconds was targeted for patients with an average risk of bleeding, 160–180 seconds for patients with a high risk of bleeding without active bleeding, and 140–160 seconds for patients with active ongoing bleeding. In patients with severe ongoing bleeding the heparin infusion was withheld until bleeding was controlled.

The anti-Xa target protocol consisted of targeting an anti-Xa level of 0.5–0.7 IU/ml. The anti-Xa level was measured twice a day and more often at the clinician’s discretion. Anti-Xa levels were measured on the patient’s native plasma sample without the addition of exogenous plasma or antithrombin, and therefore reflect a functional heparin activity based on combined heparin and antithrombin concentrations. An individualized patient ACT goal was used to titrate the heparin infusion based on the ACT obtained at the time of the last concurrent anti-Xa level. If the anti-Xa level was within the goal range, heparin was titrated to an ACT within 20 seconds above or below the ACT value obtained at the same time as the anti-Xa level. If the anti-Xa level was higher than the goal, a lower ACT was targeted and vice versa. Activated clotting time measurements were done hourly initially and decreased to every 4 hours as the patient stabilized. The ACT goal was reevaluated with every subsequent anti-Xa level. Bleeding risk assessment was not used to determine the ACT goal. Frequent titrations of the heparin infusion and boluses of heparin were discouraged. The heparin infusion was withheld similarly to the ACT targeted protocol for severe ongoing bleeding.

Antithrombin levels were routinely measured throughout the entire study period. Antithrombin concentrate was used at the treating clinicians’ discretion, generally for a low measured antithrombin level (less than 70%) accompanied with a subtherapeutic ACT or anti-Xa level despite a normal or high dose of heparin administered (greater than 20 units/kg/hour in children more than a year of age or greater than 28 units/kg/hour in children less than a year of age).

Data Collection

Patient data were collected for each ECMO run. A single patient could have multiple runs, the start and stop points of which were identified by removal of cannulas. Trials of ECMO support without removal of cannulas and subsequent resumption of ECMO support were considered the same ECMO run.

Demographic variables included age, neonate (less than 30 days of age) classification, weight, survival to discharge, hours of ECMO support, and whether the patient had recent cardiac surgery (within 48 hours before ECMO cannulation).

The primary hemorrhagic outcomes were surgical exploration while on ECMO and intracranial hemorrhage. Time to surgical exploration and time to identification of intracranial hemorrhage were recorded. Secondary indicators of hemorrhagic complications include total blood loss (through chest tube and weighed dressings), volume of packed red blood cell (pRBC) transfusion, volume of fresh frozen plasma (FFP) transfusion, volume of platelet transfusion, and the use of recombinant factor VIIa (rFVIIa). Blood loss and transfusion volumes were recorded per day for the first week of ECMO support.

The primary thrombotic outcome was circuit change. A circuit change was defined as the exchange of the entire or any component of the circuit. The time to first circuit change was recorded as well as the total number of circuit changes. Secondary thrombotic outcome was thrombosis in the patient defined as a thrombosis identified by radiological imaging test obtained for any reason from after ECMO cannulation through hospital discharge.

The effect the anticoagulation protocol change had on dosing of heparin was examined. The heparin dose was recorded as the average hourly infusion rate over the first 24 hours and first week of ECMO support. The number of heparin bolus doses, number of heparin dose changes, proportion of time heparin was on hold, hours with an ACT value less than 160 seconds, and any use of antithrombin concentrate were recorded over the entire length of ECMO support.

Data Analysis

Demographic and treatment variables were summarized as count and percent if categorical, as mean ± standard deviation if normally continuous, and as median and interquartile range if skewed continuous. Differences in the incidences of surgical exploration, intracranial hemorrhage, circuit change, and survival between treatment cohorts were compared with χ2 tests for cumulative incidences, and by Wilcoxon log-rank tests and Cox proportional hazard models for time-varying incidences. Logistic regression and Mann–Whitney test were used to model the categorical and continuous outcomes respectively over all the patients. For the outcomes spanning more than 1 ECMO run per patient, panel regression techniques were employed. Calculations were performed using Stata (StataCorp. 2015. Stata Statistical Software: Release 14; StataCorp LP, College Station, TX).


The ACT target group included 152 ECMO runs in 129 patients totaling 634.6 days on ECMO. The anti-Xa target group included 122 ECMO runs in 101 patients totaling 868.1 days on ECMO. Table 1 shows the demographics of the study populations. Patient age, weight, time between ECMO runs (in patients with multiple runs), and survival to discharge were similar in both groups. The anti-Xa target group was supported on ECMO for a longer period of time and was less likely to have had recent cardiac surgery than the ACT target group. Although the overall survival to discharge was similar in the groups, the probability of survival to hospital discharge by duration of ECMO support showed a significant advantage to the anti-Xa target group (Figure 1).

Table 1
Table 1:
Demographics of Study Cohort by Anticoagulation Strategy Group
Figure 1
Figure 1:
Kaplan–Meier plots of point estimates and 95% confidence intervals for probability of survival to hospital discharge by duration of extracorporeal membrane oxygenation (ECMO) support. Hazard ratio is shown for antifactor Xa (anti-Xa) target vs. activated clotting time (ACT) target groups in Cox model.

Surgical explorations occurred 191 times over the 634.6 days of ECMO support (3.3 ECMO days/surgical exploration) and 111 times over the 868.1 days of ECMO support (7.8 ECMO days/surgical exploration) in the ACT target and anti-Xa target groups, respectively. Time to first surgical exploration was significantly longer in the anti-Xa target group (Figure 2, A; hazard ratio: 0.68; 95% confidence interval [CI]: 0.57–0.82; p < 0.001). When restricted to patients with cardiac surgery within 48 hours of ECMO initiation the risk of surgical exploration was still reduced (Figure 2, B; hazard ratio: 0.77; 95% CI: 0.62–0.96; p = 0.022). The incidence (Table 2) and time to identification of intracranial hemorrhage (Figure 3) was significantly longer in the anti-Xa target group (hazard ratio: 0.62; 95% CI: 0.46–0.85; p = 0.002). Blood loss, pRBC, FFP, platelet volumes transfused, and the administration of rFVIIa were significantly lower in the anti-Xa target group (Table 2).

Table 2
Table 2:
Hemorrhagic Outcomes by Anticoagulation Strategy Group
Figure 2
Figure 2:
Kaplan–Meier plots of point estimates and 95% confidence intervals for time to first surgical exploration in the (A) entire cohort by study group and (B) in the patients who had been initiated on extracorporeal membrane oxygenation (ECMO) within 48 hours of a cardiac surgery. Hazard ratio is shown for antifactor Xa (anti-Xa) target vs. activated clotting time (ACT) target groups in Cox model.
Figure 3
Figure 3:
Kaplan–Meier plot of point estimates and 95% confidence intervals for time to identification of intracranial hemorrhage. Hazard ratio is shown for antifactor Xa (anti-Xa) target vs. activated clotting time (ACT) target groups in Cox model.

Extracorporeal membrane oxygenation circuit changes occurred 62 times over the 634.6 days of ECMO support (10.2 ECMO days/circuit change) and 46 times over the 868.1 days of ECMO support (18.9 ECMO days/circuit change) in the ACT target and anti-Xa target groups, respectively. Time to first circuit change was significantly longer in the anti-Xa target group (Figure 4, A; hazard ratio: 0.66; 95% CI: 0.51–0.84; p < 0.001). As recent cardiac surgery could be a significant confounder for need for a circuit change, we also compared the groups across this variable and found no significant difference (Figure 4, B; hazard ratio: 0.85; 95% CI: 0.59–1.21; p = 0.359). Thrombosis in the patient showed no significant difference between the 2 groups with 9 of 129 (7.0%) and 9 of 101 (8.9%) in the ACT target and anti-Xa target groups respectively (p = 0.588).

Figure 4
Figure 4:
Kaplan–Meier plots of point estimates and 95% confidence intervals for time to first circuit change in the (A) entire cohort by study group and (B) in the patients who had been initiated on extracorporeal membrane oxygenation (ECMO) within 48 hours of a cardiac surgery. Hazard ratio is shown for antifactor Xa (Anti-Xa) target vs. activated clotting time (ACT) target groups in Cox model.

Heparin dosing on the first day and over the first week was similar in the two groups. The heparin infusion was held for a similar proportion of time in both groups. The proportion of patients who received antithrombin concentrate was similar in both groups. The anti-Xa target group had significantly fewer heparin dose changes, bolus heparin doses, and a lower proportion of time with an ACT less than 160 seconds (Table 3).

Table 3
Table 3:
Heparin and Antithrombin Dosing by Anticoagulation Strategy Group


The change in anticoagulation protocol and circuit technology at our institution is associated with a significant decrease in hemorrhagic complications including intracranial hemorrhage, as well as a decrease in the need for circuit change.

Prior studies have found no correlation between ACT levels and heparin dosing, but have demonstrated a significant relationship between anti-Xa levels and heparin dose.8,9 In a retrospective study of 604 pediatric patients on ECMO, Baird et al.9 found no association between maintenance of ACT within recommended range and overall survival. There was increased survival seen in patients whose heparin dose was increased regardless of ACT. Others have investigated the use of anti-Xa levels for heparin titration in pediatric ECMO patients. O’Meara et al.17 showed the feasibility of frequent anti-Xa measurements without increasing blood draws or complications. Northrop et al.18 in a study similar to ours, demonstrated with the use of a thromboelastography and anti-Xa levels, a decrease in hemorrhagic complications (cannula site and surgical site bleeding), decrease in transfusion requirements, and longer circuit life. Although our results help to corroborate those by Northrop et al., our findings go 1 step further by demonstrating even more concrete clinical outcomes in terms of probability of survival, need for surgical exploration, the incidence of and time to discovery of intracranial hemorrhage, and circuit lifespan in addition to transfusion requirements.

The heparin infusion rate administered in our study populations was similar before and after the protocol change. The largest difference in heparin administration was demonstrated in the decrease in frequency of bolus doses and infusion dosing changes. This change was prescribed in the protocol and associated with a significantly less amount of time spent with an ACT less than 160 seconds. We interpret this finding as the protocol change effectively eliminated wide swings in too much and too little heparin dosing to our patients. Perhaps the decrease in bleeding and thrombotic complications seen is because of the reduction in these wide swings.

Because of the retrospective nature of the study, there were limitations in controlling for some confounding variables. Aligned in time with the change in protocol was also a change in circuit composition, from a mix of roller and centrifugal pumps with a heparin bonded circuit to a single newer generation centrifugal model and oxygenator with an albumin/heparin bonded circuit. Studies have shown increased hemolysis and renal complications with patients on centrifugal compared with roller pumps.13,14,19 No publication has demonstrated a difference in survival or hemorrhagic/thrombotic complication rates with a change in pump technology, oxygenator, or surface coating. As the change in technology coincided with the change in the anticoagulation protocol in our study sample, it is impossible to separate the effects from each other, and the results presented are representative of both a change in circuit technology and anticoagulation protocol.

Another inherent limitation to a retrospective study design is we are limited to the information contained within the medical record and cannot go back to probe the minds of the treating clinicians. In the ACT target protocol, the ACT goal was based on the assessment of bleeding risk. Unfortunately, this assessment was not recorded in the medical record. We have no ability to discern what the ACT goal was for any individual patient at a specific time and therefore have no ability to determine the amount of time spent within the goal range. Although this is a limitation to the data we can present, the results represent the application of the protocol in a real-world environment outside of the controlled confines of a prospective clinical trial where monitors are in place to keep clinicians within the boundaries of the protocol. As ACT measurements and heparin doses were recorded in the medical record during both protocols we compared the percentage of time the ACT was less than 160 and the heparin dose over the first week on ECMO as a means of comparing the intensity of anticoagulation between the two groups.

There was a higher proportion of patients with recent cardiac surgery in the ACT target group. This may have artificially increased the bleeding risk in this group given the recent surgical procedure. We attempted to convey the impact of this difference by analyzing the subgroups who did and did not have recent cardiac surgery. In terms of hemorrhagic complications, intracranial hemorrhagic rates were similar. Blood loss and volumes of transfused products were expectedly higher in the patients with recent cardiac surgery, but the differences between the ACT target and anti-Xa target groups were still present (not always statistically significant given the smaller numbers in the subgroup analysis). Time to first surgical exploration was still prolonged in the surgical subgroup, but most patients had a chest exploration within 20 hours in the surgical subgroup regardless of anticoagulation strategy group. There was no difference in time to first circuit change in the surgical subgroup.

Regarding accurately quantifying bleeding over time, measurements are difficult to compare due to an initial coagulopathy in all ECMO patients that begins to resolve as the patient stabilizes.20 Arnold et al.4 studied the coagulation response of neonates within their initial day on ECMO, finding a significant decrease in the activity of all clotting factors. Additionally, anticoagulation with heparin becomes less effective over time, thought to be related to depletion of antithrombin with prolonged systemic anticoagulation.2 Thus, the variable nature of antithrombin administration seen in our study cannot be excluded as a confounder.


We found an association of less bleeding and thrombotic complications following a change in technology and anticoagulation protocol in our patients supported on ECMO. Further prospective trials of an anticoagulation protocol based around measurements of anti-Xa instead of ACT are necessary to confirm our findings. Further work towards standardizing anticoagulation protocols for patients on ECMO is paramount as we attempt to further limit bleeding and thrombotic complications.


The authors would like to thank the critical care nursing staff and ECMO team at Children’s Hospital of Wisconsin for their tireless work and dedication to our patients.


1. Bowen FW, Edmunds LH Jr.. Zwischenberger JB, Steinhorn RH, Bartlett RH. Coagulation, anticoagulation, and the interaction of blood and artificial surfaces. in ECMO Extracorporeal Cardiopulmonary Support in Critical Care, 2000, pp. 2nd ed. Ann Arbor, MI, Extracorporeal Life Support Orginization, 67–96.
2. Oliver WC. Anticoagulation and coagulation management for ECMO. Semin Cardiothorac Vasc Anesth 2009.13: 154–175.
3. Muntean W. Coagulation and anticoagulation in extracorporeal membrane oxygenation. Artif Organs 1999.23: 979–983.
4. Arnold P, Jackson S, Wallis J, Smith J, Bolton D, Haynes S. Coagulation factor activity during neonatal extra-corporeal membrane oxygenation. Intensive Care Med 2001.27: 1395–1400.
5. Colby CE, Sheehan A, Benitz W, Van Meurs K, Halamek LP, Moss RL. Maintaining adequate anticoagulation on extracorporeal membrane oxygenation therapy: Hemochron Junior Low Range versus Hemochron 400. J Extra Corpor Technol 2003.35: 35–38.
6. Khaja WA, Bilen O, Lukner RB, Edwards R, Teruya J. Evaluation of heparin assay for coagulation management in newborns undergoing ECMO. Am J Clin Pathol 2010.134: 950–954.
7. Reed RC, Rutledge JC. Laboratory and clinical predictors of thrombosis and hemorrhage in 29 pediatric extracorporeal membrane oxygenation nonsurvivors. Pediatr Dev Pathol 2010.13: 385–392.
8. Nankervis CA, Preston TJ, Dysart KC, et al. Assessing heparin dosing in neonates on venoarterial extracorporeal membrane oxygenation. ASAIO J 2007.53: 111–114.
9. Baird CW, Zurakowski D, Robinson B, et al. Anticoagulation and pediatric extracorporeal membrane oxygenation: impact of activated clotting time and heparin dose on survival. Ann Thorac Surg 2007.83: 912–919; discussion 919.
10. Bembea MM, Annich G, Rycus P, Oldenburg G, Berkowitz I, Pronovost P. Variability in anticoagulation management of patients on extracorporeal membrane oxygenation: an international survey. Pediatr Crit Care Med 2013.14: e77–e84.
11. Bembea MM, Schwartz JM, Shah N, et al. Anticoagulation monitoring during pediatric extracorporeal membrane oxygenation. ASAIO J 2013.59: 63–68.
12. Graves DF, Chernin JM, Kurusz M, Zwischenberger JB. Anticoagulation practices during neonatal extracorporeal membrane oxygenation: survey results. Perfusion 1996.11: 461–466.
13. Barrett CS, Jaggers JJ, Cook EF, et al. Pediatric ECMO outcomes: comparison of centrifugal versus roller blood pumps using propensity score matching. ASAIO J 2013.59: 145–151.
14. O’Brien C, Monteagudo J, Schad C, Cheung E, Middlesworth W. Centrifugal pumps and hemolysis in pediatric extracorporeal membrane oxygenation (ECMO) patients: an analysis of Extracorporeal Life Support Organization (ELSO) registry data. J Pediatr Surg 2017.52: 975–978.
15. Silvetti S, Koster A, Pappalardo F. Do we need heparin coating for extracorporeal membrane oxygenation? New concepts and controversial positions about coating surfaces of extracorporeal circuits. Artif Organs 2015.39: 176–179.
16. Stanzel RD, Henderson M. Clinical evaluation of contemporary oxygenators. Perfusion 2016.31: 15–25.
17. O’Meara LC, Alten JA, Goldberg KG, et al. Anti-xa directed protocol for anticoagulation management in children supported with extracorporeal membrane oxygenation. ASAIO J 2015.61: 339–344.
18. Northrop MS, Sidonio RF, Phillips SE, et al. The use of an extracorporeal membrane oxygenation anticoagulation laboratory protocol is associated with decreased blood product use, decreased hemorrhagic complications, and increased circuit life. Pediatr Crit Care Med 2015.16: 66–74.
19. Barrett CS, Jaggers JJ, Cook EF, et al. Outcomes of neonates undergoing extracorporeal membrane oxygenation support using centrifugal versus roller blood pumps. Ann Thorac Surg 2012.94: 1635–1641.
20. Niebler RA, Christensen M, Berens R, Wellner H, Mikhailov T, Tweddell JS. Antithrombin replacement during extracorporeal membrane oxygenation. Artif Organs 2011.35: 1024–1028.

extracorporeal membrane oxygenation; anticoagulants; pediatrics; hemorrhage; thrombosis; blood coagulation tests

Copyright © 2019 by the American Society for Artificial Internal Organs