Content uploaded by Patty Theurer
Author content
All content in this area was uploaded by Patty Theurer on Jan 17, 2020
Content may be subject to copyright.
Greater Volume of Acute Normovolemic
Hemodilution May Aid in Reducing
Blood Transfusions After Cardiac Surgery
Joshua Goldberg, MD, Theron A. Paugh, CCP, Timothy A. Dickinson, MS,
John Fuller, CCP, Gaetano Paone, MD, Patty F. Theurer, BSN, Kenneth G. Shann, CCP,
Thoralf M. Sundt, III, MD, Richard L. Prager, MD, and Donald S. Likosky, PhD, for the
PERForm Registry and the Michigan Society of Thoracic and Cardiovascular Surgeons
Quality Collaborative
Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts; Department of Cardiac Surgery, University of
Michigan, Ann Arbor, Michigan; Specialty Care, Nashville, Tennessee; St. John Providence Health System Detroit Hospitals; and
Division of Cardiac Surgery, Henry Ford Hospital, Detroit, Michigan
Background. Perioperative red blood cell transfusions
(RBC) are associated with increased morbidity and mor-
tality after cardiac surgery. Acute normovolemic hemo-
dilution (ANH) is recommended to reduce perioperative
transfusions; however, supporting data are limited and
conflicting. We describe the relationship between ANH
and RBC transfusions after cardiac surgery using a multi-
center registry.
Methods. We analyzed 13,534 patients undergoing
cardiac surgery between 2010 and 2014 at any of the 26
hospitals participating in a prospective cardiovascular
perfusion database. The volume of ANH (no ANH, <400
mL, 400 to 799 mL, ‡800 mL) was recorded and linked to
each center’s surgical data. We report adjusted relative
risks reflecting the association between the use and
amount of ANH and the risk of perioperative RBC
transfusion. Results were adjusted for preoperative risk
factors, procedure, body surface area, preoperative he-
matocrit, and center.
Results. The ANH was used in 17% of the patients.
ANH was associated with a reduction in RBC trans-
fusions (RR
adj
[adjusted risk ratio] 0.74, p<0.001). Pa-
tients having 800 mL or greater of ANH had the most
profound reduction in RBC transfusions (RR
adj
0.57, p<
0.001). Platelet and plasma transfusions were also
significantly lower with ANH. The ANH population had
superior postoperative morbidity and mortality compared
with the no ANH population.
Conclusions. There is a significant association between
ANH and reduced perioperative RBC transfusion in car-
diac surgery. Transfusion reduction is most profound
with larger volumes of ANH. Our findings suggest the
volume of ANH, rather than just its use, may be an
important feature of a center’s blood conservation
strategy.
(Ann Thorac Surg 2015;100:1581–7)
Ó2015 by The Society of Thoracic Surgeons
Numerous publications have demonstrated an asso-
ciation between perioperative red blood cell trans-
fusions (RBC) and higher risk of morbidity (eg, renal
failure, respiratory failure, stroke, infections) and mor-
tality after cardiac surgery [1–3]. Indeed, single and multi-
center studies have demonstrated safety and decreased
morbidity and mortality associated with blood conserva-
tion measures [4]. As a result, numerous blood conser-
vation strategies are recommended, including acute
normovolemic hemodilution (ANH) [4].
As practiced in cardiac surgery, ANH is the process by
which whole blood is removed, collected, and stored from
a patient after induction of anesthesia and prior to hep-
arinization for cardiopulmonary bypass (CPB). The ANH
volume is replaced with sufficient volumes of colloid or
crystalloid solutions to maintain hemodynamic stability.
The autologous stored whole blood serves as a “blood
bank”for the patient to receive non-diluted, fresh, whole
blood containing red blood cells and critical clotting fac-
tors. Despite its theoretical benefits in reducing RBC
transfusions, single center reports and meta-analyses
have demonstrated mixed results regarding the effec-
tiveness of ANH to reduce RBC transfusions [5, 6–12].In
its most recent blood management guidelines, the Society
Accepted for publication April 23, 2015.
Presented at the Fifty-first Annual Meeting of The Society of Thoracic
Surgeons, San Diego, CA, Jan 24–28, 2015.
Address correspondence to Dr Likosky, Section of Health Services
Research and Quality, Department of Cardiac Surgery, University of
Michigan Medical School, Ann Arbor, MI 48109; e-mail: likosky@med.
umich.edu.
Dr Sundt discloses a financial relationship with
Thrasos Therapeutics.
Ó2015 by The Society of Thoracic Surgeons 0003-4975/$36.00
Published by Elsevier http://dx.doi.org/10.1016/j.athoracsur.2015.04.135
ADULT CARDIAC
of Thoracic Surgeons and Society of Cardiovascular An-
esthesiologists endorse ANH as a potential mechanism
for blood conservation, but acknowledge the disparate
data supporting its practice [4].
We undertook a multi-center, observational study to
identify the association between ANH use and RBC
transfusions among patients undergoing cardiac surgery
using a voluntary, multi-institutional registry of merged
perfusion and cardiac surgical data. We hypothesized
that patients exposed to ANH (and increased volume of
ANH) would have lower rates of RBC transfusions.
Patients and Methods
Patient Population
This study was approved by the Institutional Review
Board (IRB) of the University of Michigan Health System
(IRB HUM00053934, Notice of Determination of “Not
Regulated”Status). The PERFusion measures and out-
comes (PERForm) registry was established in 2010 as a
voluntary database. Current efforts are focused on iden-
tifying perfusion practices associated with improved
outcomes and providing benchmarking opportunities to
support local and multi-institutional quality improve-
ment initiatives. It is organizationally structured within
the Michigan Society of Thoracic and Cardiovascular
Surgeons Quality Collaborative (MSTCVS-QC). At the
time of this publication, 27 of 33 hospitals participating in
the MSTCVS-QC contributed data to the PERForm reg-
istry, with an additional 8 centers located outside of
Michigan [7, 8]. The MSTCVS-QC began in 2001 as a
cardiac surgeon-led quality collaborative embedded in
the Michigan Society of Thoracic and Cardiovascular
Surgeons, and in 2005 it became partially funded by the
Blue Cross/Blue Shield of Michigan. The Collaborative
meets quarterly to review various processes and out-
comes and to facilitate and evaluate quality improvement
studies.
All programs in the MSTCVS-QC utilize the Society of
Thoracic Surgeons (STS) data collection form and sub-
mit data on a quarterly basis to both the STS database
and the MSTCVS-QC data warehouse. The PERForm
registry contains information related to the care and
conduct of cardiovascular perfusion practices (a list of
fields and definitions may be found at http://www.
mstcvsqualitycollaborative.org/perform-registry.) Each
surgical record is merged with a record from the
PERForm registry [9]. Participating sites are routinely
audited for data validity and accuracy as part of the
MSTCVS-QC audit system.
We included all patients 18 or greater years of age
operated on at one of 27 participating medical centers
between the second quarter of 2010 through the second
quarter of 2014 who underwent one of the following op-
erations: isolated coronary artery bypass grafting (CABG),
isolated aortic valve (AVR), AVRþCABG, mitral valve
(MV) repair, MV replacement, CABGþMV repair or
replacement, totaling 19,434 patients. Our final dataset
included 13,534 patients; we excluded patients whose
procedures do not have a STS predicted risk of mortality
(STS PROM), those with missing information on ANH
use, patients undergoing off-pump surgery, or surgery for
endocarditis. The ANH use and volume removed was
dictated by clinicians at each institution.
Measures
The primary outcome for this analysis was any periop-
erative RBC transfusion defined as occurring during the
intraoperative or postoperative stay. We additionally
report crude rates of plasma and platelet transfusions as
well as 30-day mortality, re-operation for bleeding, post-
operative stroke, renal dysfunction and failure, and pro-
longed (>12 day) length of stay. Indications for red blood
cell transfusions were determined by the clinical team
caring for the patient. There was not a uniform trans-
fusion protocol across all participating centers. We define
ANH as the practice of removing autologous whole blood
prior to systemic heparinization for CPB with the express
purpose of reinfusion after protamine reversal. Hemo-
dynamic stability was maintained by vasopressor support
and the administration of sufficient volumes of colloid or
crystalloid solutions per institutional protocol. We cate-
gorized the volume of harvested ANH based on the
estimated volume contained in an autologous blood
collection bag (ie, <400 mL ¼1 bag, 400 to 799 ¼2 bags,
800 ¼3þbags).
Statistical Analyses
Standard statistical tests were used, including
c
2
tests for
categoric data and 2-sided Wilcoxon rank sum tests for
non-normally distributed continuous variables. Trends in
patient characteristics, processes of care, and clinical
outcomes were tested using non-parametric tests of
trend.
Analysis of the ANH cohort was divided into 4 groups:
all ANH recipients, those who had less than 400 mL
removed, those who had 400 to 799 mL removed, and
those who had 800 mL or greater removed. We explored
indexing ANH volume removed to patient body surface
area (BSA), body mass index, and estimated blood vol-
ume, but found no impact on results; thus, we chose to
stratify simply by volume removed. Baseline de-
mographic variables were reported on all patients, as
well as preoperative comorbidities. We report crude and
adjusted relative risks for RBC transfusions using Pois-
son regression. We adjusted for each patient’sSTS
PROM, BSA, preoperative hematocrit, medical center,
net prime volume, cell salvage utilization, cell salvage
volume, and procedure. Adjusted rates for postoperative
outcomes are also reported. Statistical analyses were
performed using Stata 13.0 (StataCorp, College Station,
TX). The tests were considered significant at a pvalue
less than 0.05.
Results
A total of 13,534 patients were analyzed, of whom 2,337
(17%) underwent ANH. Among ANH patients, 308
(13.2%) had less than 400 mL removed, 958 (41.9%) had
1582 GOLDBERG ET AL Ann Thorac Surg
ACUTE NORMOVOLEMIC HEMODILUTION 2015;100:1581–7
ADULT CARDIAC
Table 1. Preoperative Characteristics by Use and Volume of Acute Normovolemic Hemodilution (ANH)
Variables No ANH ANH
ANH Volume
pValue
a
pValue
b
<400 mL 400–799 mL 800 mL
Observations 11,197 2,337 308 958 1,071
Patient age (%) 83% 17% 2% 7% 8%
<60 years 27.2 33.5 34.1 29.4 37.0
60–69 years 32.6 32.5 27.9 35.0 31.7
70þyears 40.2 34.0 38.0 35.6 31.4 <0.001 <0.001
Female (%) 31.0 24.9 23.1 27.2 23.3 <0.001 <0.001
Body surface area (m
2
) (%)
<1.6 4.7 4.4 3.6 6.7 2.5
1.6–1.79 14.4 11.9 13.0 14.6 9.1
1.8–1.99 25.2 25.1 27.0 23.6 26.0
2þ55.7 58.7 56.5 55.1 62.5 <0.001 <0.001
Estimated blood volume (L)
Mean 4.7 4.8 4.8 4.7 4.9 <0.001 <0.001
Median 4.7 4.8 4.8 4.7 4.9 <0.001 <0.001
Hct, last preoperative (%)
<36 30.6 16.2 19.0 20.9 11.1
36–39 28.8 29.0 32.1 30.3 26.8
40–42 24.0 32.1 28.6 28.1 36.9
43þ16.5 22.7 20.3 20.6 25.3 <0.001 <0.001
Predicted risk of mortality
Median 1.4% 1.1% 1.1% 1.2% 0.9% <0.001 <0.001
<1% 37.5 47.7 43.8 43.8 52.2
1–1.9% 23.9 23.5 26.0 23.1 23.3
2–2.9% 12.3 11.7 9.4 12.9 11.2
3%þ26.4 17.1 20.8 20.2 13.4 <0.001 <0.001
Comorbid disease (%)
Myocardial infarction
None 58.6 72.53 65.6 67.8 78.8
Within 7 days 19.2 9.2 12.7 13.1 4.8
7 days 22.2 18.3 21.8 19.2 16.4 <0.001 <0.001
CHF 22.3 17.0 18.2 19.7 14.3 <0.001 <0.001
NYHA class III/IV 16.8 11.7 14.3 14.2 8.8 <0.001 <0.001
Aortic stenosis, severe 13.5 17.9 16.6 15.6 20.4 <0.001 <0.001
Diabetes mellitus (%) 41.0 32.1 40.3 37.0 25.5 <0.001 <0.001
Hemoglobin A1c (median) 5.9 5.9 5.9 6.0 5.8 0.17 <0.001
Peripheral arterial disease (%) 14.9 9.0 14.0 11.9 5.0 <0.001 <0.001
EGFR
90 43.8 49.6 51.6 46.6 51.6
60–89 33.5 33.1 30.5 32.7 34.2
30–59 18.6 15.1 15.9 17.8 12.4
<30 or kidney failure 4.1 2.3 2.0 2.9 1.8 <0.001 <0.001
Ejection fraction
<0.40 15.1 11.1 16.3 10.9 9.9
0.40–0.49 14.2 9.1 12.8 10.1 7.2
0.50-0.59 31.8 25.0 33.0 29.3 19.0
0.60þ39.0 54.7 37.9 49.7 63.9 <0.001 <0.001
Previous operation (%) 6.3 10.4 3.9 9.0 13.6 <0.001 <0.001
Urgent/emergent (%) 50.0 31.8 47.1 38.9 21.0 <0.001 <0.001
a
For the comparison of ANH use.
b
For the comparison across ANH volume categories.
CHF ¼congestive heart failure; EGFR ¼estimated glomerular filtration rate; NYHA ¼New York Heart Association.
1583Ann Thorac Surg GOLDBERG ET AL
2015;100:1581–7 ACUTE NORMOVOLEMIC HEMODILUTION
ADULT CARDIAC
400 to 799 mL removed, and 1,071 (45.8%) had 800 mL or
greater removed. The ANH was utilized in 21 of the 26
(81%) centers contributing data to the database with most
centers (62%) using it in less than 20% of cases, 4 (16%)
using it between 20% and 59% of cases, and 4 (16%) using
it in 60% or greater of cases. The ANH was employed in
all types of identified procedures (isolated CABG, AVR-
CABG, MV repair-CABG, MV replacement-CABG,
AVR, MV repair, and MV replacement) regardless of
center level of usage.
Table 1 details baseline demographic descriptors and
preoperative comorbidities of the studied patients. There
are numerous differences between the ANH and non-
ANH populations. Overall, ANH patients tend to be
younger, male, and have a larger BSA. Furthermore,
ANH patients tend to be less anemic with slightly higher
estimated blood volumes. The ANH patients had fewer
comorbidities reflected in a lower STS risk of mortality
(1.1% vs 1.4%, p<0.001). When stratified by ANH volume
removed, there are numerous noteworthy differences
between cohorts. Compared with patients who had
smaller volumes removed, patients who had 800 mL or
greater removed tended to be younger, male, and have a
larger BSA with higher baseline hematocrit (Hct) and
larger estimated blood volumes. With the exception of
severe aortic stenosis, patients who had ANH volumes
800 mL or greater had lower comorbidity profiles and a
lower STS PROM. Severe aortic stenosis was significantly
higher in the ANH population with the 800 mL or greater
cohort.
Operative data from the analyzed patients are detailed
in Table 2. Retrograde autologous priming (RAP) of the
CPB circuit was used more frequently (82.8% vs 71.4%,
p<0.001) and at greater volumes in ANH patients. Over
60% of patients who had 800 mL or greater of ANH vol-
ume had 500 mL or greater of RAP volume. As a conse-
quence of increased RAP among ANH patients, smaller
crystalloid CPB prime volumes were used as well. Over-
all, ANH patients received more crystalloid volume, with
increasing volume administered to patients with greater
ANH volume removed. There was no difference in nadir
Hct between ANH and no-ANH patients, including those
patients with a nadir Hct less than 21 (p¼0.5). The ANH
patients received less cell salvage volume (455 mL vs 658
mL, p<0.001).
We collected information concerning the first post-
operative Hct in the intensive care unit (ICU). The mean
first postoperative Hct in the ICU was not appreciably
different by ANH use; 30.9 (ANH) versus 30.5 (no ANH),
pless than 0.001. The mean first postoperative Hct in the
Table 2. Intraoperative Practices by Use and Volume of Acute Normovolemic Hemodilution (ANH)
Variables No ANH ANH
ANH Volume
pValue
a
pValue
b
<400 mL 400–799 mL 800 mL
Observations 11,197 2,337 308 958 1,071
CPB time (median) 97 111 116 112.5 108 <0.001 <0.001
Clamp time (median) 74 85 87 87 81 <0.001 <0.001
Retrograde autologous prime (%)
No RAP 28.6 17.2 14.0 12.6 22.3
<500 mL 20.7 20.2 14.9 30.4 12.5
500–699 mL 26.5 21.1 9.1 22.7 23.2
700 mL 24.2 41.5 62.0 34.3 42.0 <0.001 <0.001
Net prime volume (%)
500 mL 14.8 26.1 52.6 29.8 15.2
500–999 mL 41.8 46.4 40.9 46.7 47.8
1–1.49 mL 23.8 20.3 5.2 17.6 27.1
1.5 L 19.5 7.1 1.3 6.0 9.9 <0.001 <0.001
Net fluid volume administered on bypass (%)
<500 mL 46.4 43.9 61.1 46.1 37.1
500–999 mL 21.5 22.0 21.3 23.4 21.0
1 L 32.1 34.1 17.6 30.6 41.9 0.02 0.00
Nadir Hct (%)
<21 16.6 14.2 15.3 16.6 11.5
21–23 20.6 23.7 30.6 23.1 22.0
24–25 13.8 17.8 15.6 19.1 17.3
26þ49.0 44.4 38.4 41.2 49.1 <0.001 0.50
Cell salvage volume transfused (mL), median 658.0 455.0 405.0 440.0 500.0 <0.001 <0.001
Ultrafiltration (%) 26.6 23.7 30.1 25.2 20.6 <0.001 <0.001
a
For the comparison of ANH use.
b
For the comparison across ANH volume categories.
CPB ¼cardiopulmonary bypass; Hct ¼hematocrit; RAP ¼retrograde autologous priming.
1584 GOLDBERG ET AL Ann Thorac Surg
ACUTE NORMOVOLEMIC HEMODILUTION 2015;100:1581–7
ADULT CARDIAC
ICU by ANH volume; 30.5 (no ANH), 30.1 (<400 mL), 30.5
(400 to 799 mL), 31.6 (800 mL), pless than 0.001.
Table 3 reports adjusted outcomes associated with
ANH utilization. The RBC transfusion rates were signifi-
cantly lower among patients in whom ANH was used
(33.5% vs 40.3%, p<0.001). Increased ANH volume was
significantly associated with progressively fewer RBC
transfusions (p<0.001), although did not impact the rate
of transfusions given solely in the operating room (p¼
0.78). The ANH use remained significantly associated
with fewer RBC transfusions (RR
adj
[adjusted risk ratio]
0.74, p<0.001), even after adjusting for preoperative risk,
procedure, BSA, preoperative Hct, net prime volume, cell
salvage utilization, cell salvage volume, and center (Fig 1).
While protective at each volume threshold, the protective
effect of RBC transfusion was most pronounced when 800
mL or greater of ANH volume is removed (RR
adj
0.57, p<
0.001). There was a significantly lower rate of plasma
(4.9% vs 8.5%, p<0.001) and platelet transfusions (5.2%
vs 8.5%, p<0.001) among those with ANH use. As with
RBC transfusions, patients having 800 mL or greater of
ANH had the lowest transfusion rates of plasma 3.3%)
and platelets (3.4%), pless than 0.001.
Risk adjusted 30-day mortality was lower among pa-
tients receiving ANH (1.5 vs 2.8%, p<0.001). Compared
with the no-ANH cohort, patients receiving ANH had less
acute kidney injury (28.2% vs 24.1%, p<0.001), and renal
failure (1.3% vs 2.9%, p<0.001). Patients receiving 800 mL
or greater of ANH had the lowest rates of acute kidney
injury (20.9%), pless than 0.001. Use of ANH was asso-
ciated with lower rates of prolonged length of stay (12.4%
vs 15.9%, p<0.001), and was lowest among patients
having 800 mL or greater of ANH, pless than 0.001. The
difference in stroke (1.3% vs 1.8%, p¼0.22) and
readmission rates (11.4% vs 12.1%, p¼0.61) associated
with ANH use did not reach statistical significance.
Comment
In our analysis of 13,534 patients in a multicenter pro-
spectively collected database we discovered a number of
findings that further support the use of ANH as an
effective blood conservation technique to prevent peri-
operative RBC transfusions during cardiac surgery.
First, the use of ANH was effective, as it was associated
with fewer allogenic red blood cell transfusions, even
after adjustment for pre-operative risk factors, medical
center and procedure type. Second, ANH use was also
associated with fewer platelet and plasma transfusions.
Third, the reduction in allogeneic transfusions is most
pronounced when 800 mL or greater of ANH volume is
removed. In addition, ANH was associated with
improved risk adjusted outcomes, including prolonged
length of stay, 30-day mortality, and renal failure. The
safety of ANH combined with reduction in allogenic
blood transfusions supports its use as a component of an
overall blood conservation strategy in cardiac surgery.
Our findings support those of previous studies reporting
a reduction in allogenic transfusions with ANH utiliza-
tion. In NewYork-Presbyterian/Weill Cornell Medical
Center, 90 cardiac surgery patients were randomized to
either ANH or no ANH and a significant reduction in
RBC, platelet, and plasma transfusions was found [9].
Similarly, 100 CABG patients randomized to ANH or no
ANH by the Los Angeles Kaiser group [7] experienced a
45% reduction in allogeneic transfusions with ANH us-
age. Furthermore, a meta-analysis of randomized trials of
ANH use reported a 50% reduction in allogeneic
Table 3. Adjusted Postoperative Outcomes by Use and Volume of Acute Normovolemic Hemodilution (ANH)
Variables No ANH ANH
ANH Volume
pValue
a
pValue
b
<400 mL 400–799 mL 800 mL
Observations 11,197 2,337 308 958 1,071
Reoperation for bleeding (%) 2.3% 1.9% 2.5% 2.0% 1.0% 0.30 0.09
Stroke (%) 1.8% 1.3% 1.8% 1.4% 1.0% 0.22 0.14
Acute kidney injury (%) 28.2% 24.1% 28.3% 25.6% 20.9% <0.001 <0.001
Renal failure (%) 2.9% 1.3% 2.8% 1.4% 1.1% <0.001 <0.001
Intraaortic balloon pump (%) 8.2% 4.9% 7.5% 5.3% 3.3% <0.001 <0.001
Red blood cells (%)
None 59.7% 66.5% 66.1% 63.1% 70.7% <0.001 <0.001
Intraoperative only 8.5% 7.7% 3.5% 9.4% 7.1% 0.78 0.81
Postoperative only 21.9% 16.1% 25.9% 18.7% 6.6% <0.001 <0.001
Intraoperative þpostoperative 10.4% 4.9% 5.9% 6.6% 2.7% <0.001 <0.001
Plasma (%) 8.5% 4.9% 7.3% 5.8% 3.3% <0.001 <0.001
Platelets (%) 8.5% 5.2% 7.8% 6.1% 3.4% <0.001 <0.001
Prolonged length of stay (%) 15.9% 12.4% 18.2% 13.0% 10.5% <0.001 <0.001
30-day mortality 2.8% 1.5% 2.0% 0.01% 2.0% <0.001 0.01
Readmission (%) 12.1% 11.4% 9.5% 12.5% 10.5% 0.61 0.66
a
For the comparison of ANH use.
b
For the comparison across ANH volume categories.
Adjusted for preoperative hematocrit, body surface area, net prime volume, cell saving device volume, center and procedure.
1585Ann Thorac Surg GOLDBERG ET AL
2015;100:1581–7 ACUTE NORMOVOLEMIC HEMODILUTION
ADULT CARDIAC
transfusions in cardiac surgery patients [11]. Our data
conflict with previous reports that have found no signif-
icant effect of ANH on postoperative RBC transfusions,
including randomized trials by Casati and colleagues (n ¼
200) [13] and Hohn and colleagues (n ¼80) [10]; both of
these studies showed no significant reduction in alloge-
neic transfusions. While cardiac surgery patients were not
analyzed separately, a meta-analysis of cardiac and
vascular surgery patients found no reduction in trans-
fusions with the use of ANH [12].
Our study highlights a potentially critical point in the
practice of ANH; the volume removed matters. We found
the most profound reduction in transfusions occurred
when 800 mL or greater of ANH volume is removed.
Similarly, the Bryson meta-analysis [11] observed the
most significant blood conservation impact occurred
when ANH volumes exceeded 1 L, while Helm and col-
leagues [9] and Kochamba and colleagues [7] found a
reduction in transfusions with removal of mean ANH
volumes of 1,540 mL and 866 mL, respectively. Further-
more, studies demonstrating a null effect tended
to remove smaller volumes of ANH [10, 13]. The null
findings of these studies may be reflective of a
sub-therapeutic volume of ANH rather than the lack of
utility of ANH in general. Thus, ANH may be not only
underutilized in frequency but also used insufficiently
when employed. In addition, our data demonstrate a
significant portion of patients who undergo ANH also
undergo RAP, with the highest proportion in the 800 mL
or greater cohort (60% of which had 500 mL of RAP
volume), which implies appropriately selected patients
can tolerate relatively large shifts of their blood volume.
While further research is needed to determine the dose–
response relationship of ANH on transfusion prevention,
our large multi-institutional study shows that ANH is not
only safe and effective at reducing RBC transfusions, but
is most effective when removing 800 mL or greater of a
patient’s blood volume. We recognize some limitations to
our current study. First, we cannot rule out the effect of
unmeasured confounding, including other institutional or
physician-related practices [14, 15]. We employed stan-
dard approaches, including risk adjustment, to address
apparent differences in preoperative characteristics. Sec-
ond, we recognize that there are a number of clinical
reasons and patient level factors that impact the clinical
decision to use ANH. With that being said, we could not
find evidence of adverse harm associated with the use of
ANH, including among patients with ejection fraction less
than 0.40, those undergoing urgent or emergent opera-
tions, and those with STS predicted mortality risk of 3%
or greater.
This contemporaneous, multi-institutional study
demonstrates the reduction in transfusions associated
with the use of ANH. Furthermore, our findings suggest
that a therapeutic ANH volume may be higher than that
removed by many centers. While additional studies are
needed to identify the optimal volume of ANH to
remove in a given patient, clinicians should consider
employing ANH as part of a larger blood management
strategy.
This study was approved by the Institutional Review Board (IRB)
of the University of Michigan Health System (IRB HUM00053934,
Notice of Determination of “Not Regulated”Status). The
MSTCVS Quality Collaborative recognizes the support of Blue
Cross Blue Shield of Michigan and Blue Care Network.
Dr Likosky is supported in part by the following grants from the
Agency for Healthcare Research and Quality (AHRQ) and U.S.
Department of Health and Human Services: R01HS022535 and
R03HS022909. The opinions expressed in this document are
those of the authors and do not reflect the official position of
AHRQ or the U.S. Department of Health and Human Services.
We acknowledge the editorial review provided by Amanda
Schuetz and Katie Wopinsky.
References
1. Murphy GJ, Reeves BC, Rogers CA, Rizvi SI, Culliford L,
Angelini GD. Increased mortality, postoperative morbidity,
and cost after red blood cell transfusion in patients having
cardiac surgery. Circulation 2007;116:2544–52.
2. Koch CG, Li L, Duncan AI, et al. Morbidity and mortality risk
associated with red blood cell and blood-component trans-
fusion in isolated coronary artery bypass grafting. Crit Care
Med 2006;34:1608–16.
3. Surgenor SD, DeFoe GR, Fillinger MP, et al. Intraoperative
red blood cell transfusion during coronary artery bypass
graft surgery increases the risk of postoperative low-output
heart failure. Circulation 2006;114(1 Suppl):I43–8.
4. Ferraris VA, Brown JR, Despotis GJ, et al. 2011 update to the
Society of Thoracic Surgeons and the Society of Cardiovas-
cular Anesthesiologists blood conservation clinical practice
guidelines. Ann Thorac Surg 2011;91:944–82.
5. Messmer K, Lewis DH, Sunder-Plassmann L, Klovekorn WP,
Mendler N, Holper K. Acute normovolemic hemodilution.
Changes of central hemodynamics and microcirculatory flow
in skeletal muscle. European Surg Res 1972;4:55–70.
6. von Heymann C, Sander M, Foer A, et al. The impact of an
hematocrit of 20% during normothermic cardiopulmonary
bypass for elective low risk coronary artery bypass graft
surgery on oxygen delivery and clinical outcome–a ran-
domized controlled study [ISRCTN35655335]. Crit Care
2006;10:R58.
7. Kochamba GS, Pfeffer TA, Sintek CF, Khonsari S. Intra-
operative autotransfusion reduces blood loss after cardio-
pulmonary bypass. Ann Thorac Surg 1996;61:900–3.
Fig 1. Rate of red blood cell transfusions by volume of acute nor-
movolemic hemodilution removed. Relative risks (RRadj) are also
presented after adjustment for each patient’s The Society of Thoracic
Surgeons Predicted Risk of Mortality, body surface area, preoperative
hematocrit, medical center, and procedure.
1586 GOLDBERG ET AL Ann Thorac Surg
ACUTE NORMOVOLEMIC HEMODILUTION 2015;100:1581–7
ADULT CARDIAC
8. Licker M, Ellenberger C, Sierra J, Kalangos A, Diaper J,
Morel D. Cardioprotective effects of acute normovolemic
hemodilution in patients undergoing coronary artery bypass
surgery. Chest 2005;128:838–47.
9. Helm RE, Klemperer JD, Rosengart TK, et al. Intraoperative
autologous blood donation preserves red cell mass but does
not decrease postoperative bleeding. Ann Thorac Surg
1996;62:1431–41.
10. Hohn L, Schweizer A, Licker M, Morel DR. Absence of bene-
ficial effect of acute normovolemic hemodilution combined
with aprotinin on allogeneic blood transfusion requirements
in cardiac surgery. Anesthesiology 2002;96:276–82.
11. Bryson GL, Laupacis A, Wells GA. Does acute normovolemic
hemodilution reduce perioperative allogeneic transfusion? A
meta-analysis. The International Study of Perioperative
Transfusion. Anesth Analg 1998;86:9–15.
12. Curley GF, Shehata N, Mazer CD, Hare GM, Friedrich JO.
Transfusion triggers for guiding rbc transfusion for cardio-
vascular surgery: a systematic review and meta-analysis*.
Crit Care Med 2014;42:2611–24.
13. Casati V, Speziali G, D’Alessandro C, et al. Intraoperative
low-volume acute normovolemic hemodilution in adult
open-heart surgery. Anesthesiology 2002;97:367–73.
14. Cote C, MacLeod JB, Yip AM, et al. Variation in trans-
fusion rates within a single institution: exploring
the effect of differing practice patterns on the likelihood
of blood product transfusion in patients undergoing
cardiac surgery. J Thorac Cardiovasc Surg 2015;149:
297–302.
15. Jin R, Zelinka ES, McDonald J, et al. Effect of hospital culture
on blood transfusion in cardiac procedures. Ann Thorac Surg
2013;95:1269–74.
DISCUSSION
DR ALAN M. SPEIR (Falls Church, VA): I wonder if you could
help us to understand what aortic size, in the presence of aortic
valve replacement then would you recommend leaving behind?
I raise this because many of the patients in your study were in
the 5 cm and greater range, which is clearly an indication for
replacement. We heard earlier in the first session that even at
4.7 cm, there is a 10% risk incrementally of requiring operation
with a much lower operative rate of mortality. So are you
advocating, based on your data, that a 5 cm aorta would remain
behind?
DR GOLDBERG: No. The study is not designed to really
establish a criteria of what should be replaced and what should
not and when. It is just to look at the actual risk accrued with the
addition of aortic replacement. So the denominator here is all
aortic valve replacements and not all dilated ascending aortas. It
is to look at the consequences, if you will, of strictly following the
recommendation of replacing all aortas at 4.5 cm. There are some
studies that counter some of the previously mentioned data that
show that observing mildly dilated aortas up to an eight-year
period does not accrue any increased mortality. I think it is a
gray area there. But the bottom line is that this study was not
really designed to make a recommendation on size criteria for
replacement but rather what is the risk of replacement, and in
some instances it is appropriate to take on additional risk where
in others it may not be.
DR RICHARD J. SHEMIN (Los Angeles, CA): Did you actually
look at experience of the individual surgeon?
DR GOLDBERG: That is a very good question. We did not
separate by individual surgeon. Of course, there is data that
surgeons absolutely affect outcome, and that would be an area
for another project. But I think, in a sense, if you are going to look
at the “real world”experience where things are not done by
strictly aortic surgeons, then these are more realistic outcomes.
DR SHEMIN: Also did you eliminate anyone who had a root
replacement?
DR GOLDBERG: Yes, all roots, composites, arches. The only
thing we did include was hemiarch, because we felt that it was
representative of the population that may have ascending aorta
replacement at the time of some surgeon’s choice.
DR SHEMIN: And do you know the number who had an
ascending aortic replacement with a simple interposition graft
versus the patients that needed circulatory arrest with or without
antegrade cerebral perfusion and hemiarch?
DR GOLDBERG: There were just over 20% that had a hemiarch
replacement with circ arrest, and the average circ arrest time was
about 17 minutes.
DR SHEMIN: Did that subgroup result in most of the morbidity
or mortality?
DR GOLDBERG: We did not look at that separately, but that is a
very good point that you would expect to affect the outcome.
1587Ann Thorac Surg GOLDBERG ET AL
2015;100:1581–7 ACUTE NORMOVOLEMIC HEMODILUTION
ADULT CARDIAC