ASAIO Journal 2007

Assessing Heparin Dosing in Neonates on Venoarterial

Extracorporeal Membrane Oxygenation
CRAIG A. NANKERVIS,* THOMAS J. PRESTON,* KEVIN C. DYSART,* WHITNEY D. WILKINSON,* LOUIS G. CHICOINE,*
STEPHEN E. WELTY,* AND LEIF D. NELIN*

We studied 12 consecutive neonates placed on venoarterial

extracorporeal membrane oxygenation (ECMO) in 2004
2005. Activated clotting times (ACT) and anti-factor Xa levels
were measured, and the corresponding heparin drip rate was
noted. The mean heparin drip rate was 42.2 10.9 (SD)
U/kg/hour (range 20.0 69.5 U/kg/hour). There were 55 simultaneous ACT and anti-factor Xa samples drawn. The mean
ACT was 167 20 seconds (range 128 227 seconds). There
was no correlation between ACT levels and heparin dose (r
0.21; p 0.12). The mean anti-factor Xa activities were
0.73 0.19 U/ml (range 0.1 - 1.0 U/ml). There was a correlation (r 0.75; p < 0.0001) between anti-factor Xa and
heparin dose. We also examined the effect of day on ECMO
on heparin drip rate, ACT, and anti-factor Xa. There was no
correlation between day on ECMO and either heparin drip
rate (r 0.21, p 0.12) or ACT (r 0.002, p 0.99).
However, there was a positive correlation (r 0.46, p <
0.0005) between day on ECMO and anti-factor Xa activities.
In these neonatal patients on venoarterial ECMO, ACT was
not a reliable indicator of heparin effect. Furthermore, the
increase in anti-factor Xa levels with time on ECMO suggests
that heparin accumulates and/or that anti-thrombin III levels
decrease with time on ECMO. ASAIO Journal 2007; 53:
111-114.

Figure 1. Repeated measurements in one patient over a 2-day

period using the Hemochron Jr with ACT cartridges (ACT), the
Hemochron Jr with LR cartridges (LR), or the Hemochron 801 with
celite Cartridges (801). In our hands, the Hemochron Jr with the
ACT cartridges had the greatest degree of reproducibility over
time.

patient over a 2-day period for simultaneous determinations of

ACT using each of the three cartridges. The mean ACT using
the Hemochron 801 with the celite cartridge was 222 35
(SD) seconds; using the Hemochron Jr with the LR cartridge the
mean ACT was 279 46 seconds, and using the Hemochron
Jr with the ACT cartridge the mean ACT was 183 21
seconds. The coefficient of variation for the Hemochron 801
with celite cartridge was 15.4%, for the Hemochron Jr with the
LR cartridge the coefficient of variation was 16.4%, whereas
for the Hemochron Jr with the ACT cartridge the coefficient
of variation was 11.2%.
After a period of using the new ACT machines, however, we
noticed that there was a tendency for ACT values to decrease
over time on ECMO leading to greater heparin doses than we
had previously used. One patient developed clinically significant bleeding due to lower ACT levels and resultant greater
heparin doses despite appropriate levels of clotting factors.
This led us to question whether the bedside ACT was accurately reflecting heparin effect, and thereby possibly resulting
in excessive heparin dosing. Classically, anti-factor Xa activity
is used to monitor heparin activity in a wide variety of patients.3,4,6 For neonatal ECMO, ACT is a convenient, reproducible test that requires small volumes of blood and can be done
by the bedside ECMO specialist. However, anti-factor Xa activity is not a bedside test, requires 2.7 ml of blood, has a
turnaround time of 60 minutes, and is usually only available
from the laboratory during regular work hours. Despite the
disadvantages of anti-factor Xa activity testing and given the
apparent problems with bedside ACT in these patients, we set
out to determine the relationship between heparin dose and

nticoagulation is necessary for successful neonatal venoarterial extracorporeal membrane oxygenation (ECMO), and
bedside coagulation monitoring has become a mainstay of
ECMO therapy.1,2 In our institution, we routinely use bedside
activated clotting times (ACT) to monitor and adjust heparin
dosing in patients on ECMO.3,4 Our protocol called for ACT
levels between 160 and 200 seconds. Recently, we switched
our bedside ACT machine to the Hemochron Jr (International
Technidyne, Edison, NJ) with the ACT cartridge.5 We settled
on the ACT cartridge based on more stable and repeatable
results over time in our institution compared with either the
Hemochron Jr LR cartridge or the Hemochron 801 with celite
cartridge. Figure 1 demonstrates repeated measures in one

From the *Neonatal ECMO Program, Columbus Childrens Hospital,

Columbus, Ohio, USA; Center for Perinatal Research, Columbus
Childrens Research Institute, Columbus, Ohio, USA; and Section of
Neonatology, Department of Pediatrics, The Ohio State University,
Columbus, Ohio, USA.
Submitted for consideration; accepted for publication in revised
form.
Reprint requests: Leif D. Nelin, MD, Center for Perinatal Research,
700 Childrens Drive W207, Columbus, Ohio 43205, USA.
DOI: 10.1097/01.mat.0000247777.65764.b3

111

112

NANKERVIS ET AL.

ACT levels. Anti-factor Xa activities were used to confirm

heparin effect. We tested the hypothesis that the ACT levels
will be positively correlated with heparin dose and therefore
would correlate with the anti-factor Xa activity. In this study,
each weekday on ECMO blood was obtained for bedside ACT
testing and for determination of anti-factor Xa activity, and the
heparin dose was noted at that time. Linear regression modeling was used to determine if correlations were present.
Patients and Methods
We studied patients placed on venoarterial ECMO in the
Columbus Childrens Hospital neonatal intensive care unit
during 2004 and 2005. In our institution, we use Medtronic
(Minneapolis, MN) model 0800 silicone oxygenators, non
heparin-coated circuits, and porcine heparin for neonatal
venoarterial ECMO. Heparinization was accomplished using
ELSO guidelines. ACT levels were used routinely to determine
the adequacy of anticoagulation. During the study, patients
were cared for on ECMO using the Neonatal ECMO protocols,
which are based on ELSO recommendations, and at the discretion of the attending neonatologist. We have a dedicated
neonatal ECMO team in our institution that consults with the
attending neonatologist and participates in rounds on all patients receiving venoarterial ECMO.
ACT and anti-factor Xa activity were measured simultaneously each weekday on ECMO at 1000. The corresponding heparin dose at the time of blood sampling was recorded.
Data are presented as mean SD. The correlation between
heparin dose and ACT or anti-factor Xa activity was assessed
using linear regression. The correlation between ECMO day
and heparin dose, ACT or anti-factor Xa activity was also
examined using linear regression. Linear regression was performed using Sigmaplot 9.0 (Jandel Scientific, San Rafael, CA).
Regression coefficients that yielded a p 0.05 were considered to be significant. The Institutional Review Board of the
Columbus Childrens Hospital approved this study.
Results
Seven patients were treated with venoarterial ECMO in 2004
and six were treated with venoarterial ECMO in 2005 at Columbus Childrens Hospital NICU. One patient was treated for
12 hours and did not have anti-factor Xa activities drawn, thus
data are available from 12 patients. Of these 12 patients, six
had congenital diaphragmatic hernia, two had persistent pulmonary hypertension, two had sepsis, one had pertussis, and
one had meconium aspiration syndrome. The mean patient
weight was 3.7 0.5 (SD) kg. The average length of time on
venoarterial ECMO was 274 104 hours with a range of 114
to 403 hours. The mean heparin dose was 42.2 10.9 U/kg/h
with a range of 20.0 to 69.5 U/kg/h. The mean amount of fresh
frozen plasma (FFP) given was 195 150 ml per patient. There
were a total of 55 simultaneous ACT and anti-factor Xa samples drawn from these 12 patients.
Contrary to our hypothesis, there was no correlation between ACT levels and heparin dose (r 0.21; p 0.12), as
shown in Figure 2. The mean ACT level was 167 20 seconds
with a range of 128 to 227 seconds. Conversely and as expected, anti-factor Xa activities were significantly (p 0.0001)
correlated (r 0.75) with heparin dose (Figure 3). The mean

Figure 2. Bedside ACT are not correlated with heparin dose in

neonates on venoarterial ECMO. The individual ACT values (n 55)
are plotted against the heparin dose at the time of testing. There is
no correlation between heparin dose and ACT level by linear regression, y 0.37x 182, r 0.21, p 0.12.

anti-factor Xa activity was 0.7 0.2 U/ml with a range of 0.1

to 1.0 U/ml. Because we found no correlation between ACT
and heparin dose but did find a correlation between anti-factor
Xa activity and heparin dose, we assessed the correlation
between ACT and anti-factor Xa activity but found none (r
0.14, p 0.31; Figure 4).
Because there was a wide range of time on ECMO, we also
determined if there was a correlation between the day on
ECMO and ACT or anti-factor Xa activity. We found no correlation (r 0.002; p 0.99) between ECMO day and ACT
(Figure 5). However, we did find a correlation (r 0.46, p
0.0005) between ECMO day and anti-factor Xa activity (Figure
6). To be sure that the correlation between ECMO day and
anti-factor Xa activity was not simply due to increasing heparin
requirements during the venoarterial ECMO run, we determined the correlation between day on ECMO and heparin
dose. We found that in these patients there was no such
correlation (r 0.21; p 0.12; Figure 7). The mean heparin
dose on ECMO day 1 (30.0 3.9 U/kg/hour) was lower (p
0.05) than the mean heparin dose on subsequent days (43.7
1.5 U/kg/hour), but this most likely because the heparin bolus
was administered at the initiation of venoarterial ECMO.

Figure 3. Anti-factor Xa activity is positively correlated with

heparin dose in neonates on venoarterial ECMO. The individual
anti-factor Xa activities (n 55) are plotted against the heparin dose
at the time of testing. There is a positive correlation between heparin
dose and anti-factor Xa activity by linear regression, y 0.01x
0.16, r 0.75, p 0.0001.

ACT AND ECMO

113

Figure 4. Bedside ACT is not correlated with anti-factor Xa levels

in neonates on venoarterial ECMO. The individual ACT values are
plotted against the anti-factor Xa activities for all 55 simultaneously
obtained samples. There was no correlation between ACT and
anti-factor Xa activity in these patients by linear regression, y
13.9x 176, r 0.14, p 0.31.

Figure 6. Anti-factor Xa activity is correlated with time on venoarterial ECMO. Anti-factor Xa activities plotted against ECMO day
for the 55 samples collected from 12 neonatal patients with simultaneous ACT and anti-factor Xa activity. Anti-factor Xa activity is
positively correlated with ECMO day by linear regression, y
0.024x 0.57, r 0.46, p 0.0005.

Discussion
Contrary to our hypothesis, we found that ACT levels were
not correlated with heparin dose in neonatal patients on venoarterial ECMO, demonstrating that ACT levels are not a reliable
indicator of heparin effect in these patients. As expected,
anti-factor Xa activity was related to heparin dose, demonstrating that anti-factor Xa activity is a reliable indicator of heparin
effect in neonatal patients on venoarterial ECMO. Although
anti-factor Xa activity was a reliable indicator of heparin effect,
the routine use of anti-factor Xa activity for monitoring coagulation in the ECMO setting is limited. First, measurement of
anti-factor Xa activity cannot currently be done at the bedside.
Second, anti-factor Xa activities in our institution were run
once a day, but only on weekdays, thus they are very difficult
to obtain at night or on weekends. Therefore, although antifactor Xa activity better reflects heparin effect, ACT levels
remain the best choice for bedside evaluation of coagulation in
neonatal patients on ECMO. However, our data suggest that
anti-factor Xa activity should be checked periodically in neonatal patients on ECMO to confirm appropriate heparin dosing, or when the bedside ACT levels do not appear to correlate
with the patients clinical coagulation picture.
In 1996, Graves et al.7 reported that 80% of the 81 active

neonatal ECMO centers at the time determined heparin dosages exclusively by ACT values. Although bedside ACT measurements have become the mainstay for monitoring heparin
effect in patients on ECMO, there are little data demonstrating
a correlation between ACT and heparin effect in neonates on
venoarterial ECMO. In 1990, in a series of nine infants on
venoarterial ECMO, Green et al.3 reported that ACT measurements correlated with heparin concentrations, although individual ACT results were limited by a lack of precision with the
estimate of heparin concentration from a single ACT determination having a coefficient of variation of 32%. In pediatric
patients on venoarterial ECMO, Ambrose et al.6 reported that
the correlations between activated clotting time and anti-factor
Xa activity were poor, and that an individual ACT value was
not predictive of the anti-factor Xa activity in the sample. Our
data demonstrate that, in neonates on venoarterial ECMO,
measurement of ACT levels alone do not accurately reflect
heparin effect. Taken together these data strongly support the
concept that heparin dosing in neonates on venoarterial
ECMO should be monitored by more than just the ACT values.
In our institution, simultaneous anti-factor Xa activities and
ACT values are checked routinely once in the morning on
patients on ECMO. The Table shows how we respond to a

Figure 5. Bedside ACT is not correlated with the day on venoarterial ECMO in neonates. The individual ACT levels are plotted
against ECMO day for the 55 samples collected in this study. There
is no correlation between ACT and ECMO day by linear regression,
y 0.008x 166, r 0.002, p 0.99.

Figure 7. Heparin dose in patients plotted against ECMO day for

the 55 samples collected from 12 patients with simultaneous ACT
and anti-factor Xa activity. There is no correlation between ECMO
day and heparin dose by linear regression, y 0.72x 39, r 0.21,
p 0.12.

114

NANKERVIS ET AL.

Table 1. Response to Anti-Factor Xa Activities and ACT

Values
Anti-Factor
Xa

ACT

Response

Normal

2
1
2
1
normal

1
1
2
2
2

FFP and repeat anti-Factor Xa level, if pattern

continues consider adjusting ACT range up
FFP and increase heparin drip rate
Decrease heparin drip rate
Increase heparin drip rate
FFP and decrease heparin drip rate
FFP and repeat anti-Factor Xa level, if pattern
continues consider adjusting ACT range
down

Anti-factor Xa activities and ACT values are represented as either

within normal range (normal), low (2), or high (1), since there is a
great deal of inter-institutional variability particular in the ACT values
depending on the ACT machine and cartridges used. FFP is fresh
frozen plasma and is usually administered in 10 mL/kg boluses.

given pattern of anti-factor Xa and ACT values, where antifactor Xa levels are assessed for heparin level and ACT values
are used to assess adequacy of substrate for heparin effect. As
an example, if the anti-factor Xa is high, but the ACT is low,
this would suggest that the patient requires a decreased heparin infusion rate and a bolus of FFP to provide adequate
substrate to improve the effect of the heparin on clotting. In
response to a bolus of FFP, one would expect the anti-factor Xa
activity not to change, but that the ACT values would increase
appropriately, thus, permitting the use of bedside ACT monitoring to guide anticoagulation therapy until the next antifactor Xa activity can be obtained. In addition to the once daily
anti-factor Xa activity, if the patient develops symptoms of
coagulation or bleeding in the face of normal ACT values, then
an anti-factor Xa activity is checked immediately, if possible.
Our laboratory has made provision for anti-factor Xa activity
assays to be run weekdays between 8:00 AM and 5:00 PM, and
on weekends and holidays once in the morning. Our laboratory has also made anti-factor Xa activity assays available to
patients on venoarterial ECMO on an on-call basis during off
hours with the on-call pathologist screening our request. Thus,
although not as readily available as the bedside ACT, the
anti-factor Xa activity has become an important adjunct to
bedside ACT testing in our institution to confirm ACT values
allowing for more appropriate responses to changes in the
coagulation status of the patient on venoarterial ECMO.
The increase in anti-factor Xa activity with time on venoarterial ECMO is an intriguing and clinically important finding.
On venoarterial ECMO, heparin is routinely used as the anticoagulant. Heparin inhibits thrombin and Factor Xa generation. To fully exert these effects, heparin requires antithrombin
in near-normal plasma concentrations. Thus, one potential
mechanism for the increase in anti-factor Xa activity may be
that anti-thrombin III levels decrease with time in neonates on
venoarterial ECMO. Consistent with this mechanism are the
findings of Urlesberger et al.8 that neonates on venoarterial
ECMO had a decrease in thrombin-antithrombin complex concentrations between 2 and 48 hours after the initiation of
venoarterial ECMO. Another possible mechanism for the in-

crease in anti-factor Xa activity with time on ECMO could be

that heparin accumulates in these patients. It has been shown
that heparin binds to the ECMO circuit,9 and thus heparin
clearance is high in ECMO patients.10 However, it may be that
over time the binding of heparin to the circuit decreases,
resulting in an increase in the circulating levels of heparin.
Thus, the positive correlation of anti-factor Xa activity to
ECMO day is consistent with a decrease in anti-thrombin III
levels and/or an accumulation of heparin in the neonate with
time on venoarterial ECMO.
Conclusion
In conclusion, we found that in these patients on venoarterial ECMO, ACT was not a reliable indicator of heparin effect.
Therefore, although factor Xa levels may be difficult to obtain
during off-hours, they are a reliable indicator of heparin effect
in neonates on venoarterial ECMO. Furthermore, the increase
in factor Xa levels with time on ECMO suggests that that
anti-thrombin III levels decrease with time on ECMO and/or
that heparin accumulates in these neonatal patients. Although,
ACT levels remain the best bedside check of coagulation, we
suggest that anti-factor Xa activity and ACT should be checked
simultaneously at least daily in these patients while on venoarterial ECMO to assess heparin dosing and need for additional
substrate (FFP). We speculate that the requirement for heparin
in these patients may decrease with time on venoarterial
ECMO, and that measurement of anti-thrombin III levels may
distinguish between heparin accumulation and decreased substrate for heparin effect.
References
1. Bahrami KR, Van Meurs KP: ECMO for neonatal respiratory failure. Semin Perinatol 29: 1523, 2005.
2. Zavadil DP, Stammers AH, Willett LD, et al: Hematological abnormalities in neonatal patients treated with extracorporeal
membrane oxygenation (ECMO). J Extra Corpor Technol 30:
8390, 1998.
3. Green TP, Isham-Schopf B, Steinhorn RH, et al: Whole blood
activated clotting time in infants during extracorporeal membrane oxygenation. Crit Care Med 18: 494 498, 1990.
4. Muntean W: Coagulation and anticoagulation in extracorporeal
membrane oxygenation. Artif Organs 23: 979 83, 1999.
5. Colby CE, Sheehan A, Benitz W, et al: Maintaining adequate anticoagulation on extracorporeal membrane oxygenation therapy:
Hemochron Junior Low Range versus Hemochron 400. J Extra
Corpor Technol 35: 3538, 2003.
6. Ambrose TM, Parvin CA, Mendeloff E, Luchtman-Jones L: Evaluation of the TAS analyzer and the low-range heparin management test in patients undergoing extracorporeal membrane oxygenation. Clin Chem 47: 858 866, 2001.
7. Graves DF, Chernin JM, Kurusz M, Zwischenberger JB: Anticoagulation practices during neonatal extracorporeal membrane oxygenation: survey results. Perfusion 11: 461 466, 1996.
8. Urlesberger B, Zobel G, Zenz W, et al: Activation of the clotting
system during extracorporeal membrane oxygenation in term
newborn infants. J Pediatr 129: 264 268, 1996.
9. Buck ML: Pharmacokinetic changes during extracorporeal membrane oxygenation implications for drug therapy in neonates.
Clin Pharmacokinet 42: 403 417, 2003.
10. Green TP, Isham-Schopf B, Irmiter RJ, et al: Inactivation of heparin
during extracorporeal circulation in infants. Clin Pharmacol
Ther 48: 148 154, 1990.