Vox Sanguinis (2020)

© 2020 International Society of Blood Transfusion

ORIGINAL PAPER DOI: 10.1111/vox.12994

TACO-BEL-3: a feasibility study and a retrospective audit of

diuretics for patients receiving blood transfusion at ten
hospitals
Aditi Khandelwal,1,2,3,4 Yulia Lin,1,4,5 Christine Cserti-Gazdewich,1,2,4 Muntadhar Al Moosawi,6 Chantal Armali,4,5
7,8 2,4 6,9 1,2,4
Donald Arnold, Jeannie Callum, Karen L. Dallas, Lani Lieberman, Katerina Pavenski,1,4,10
11 2,3,4,12 6,13,14
Benjamin Rioux-Masse, Nadine Shehata, Andrew W. Shih & Jacob Pendergrast1,2,4
1
Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, ON, Canada
2
Laboratory Medicine Program, University Health Network, Toronto, ON, Canada
3
Canadian Blood Services, Toronto, ON, Canada
4
Education and Safety in Transfusion (QUEST) Research Program, University of Toronto Quality in Utilization, Toronto, ON, Canada
5
Sunnybrook Health Sciences Centre, Toronto, ON, Canada
6
Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, BC, Canada
7
Hamilton General Hospital, Hamilton, ON, Canada
8
Juravinski Cancer Centre, Hamilton, ON, Canada
9
St. Paul Hospital, Vancouver, BC, Canada
10
St Michael’s Hospital, Toronto, ON, Canada
11
Centre Hospitalier de l’Universite de Montre al, Montreal, PQ, Canada
12
Mount Sinai Hospital, Toronto, ON, Canada
13
Vancouver General Hospital, Vancouver, BC, Canada
14
Centre for Blood Research, University of British Columbia, Vancouver, BC, Canada

Background and Objectives Transfusion-associated circulatory overload (TACO) is the

Received: 7 March 2020,
revised 3 July 2020,
accepted 8 August 2020
leading cause of transfusion-related morbidity and mortality. A recently completed pilot
trial randomized patients to pre-transfusion furosemide versus placebo but had a slower
than expected enrollment rate. We sought to determine whether the lack of recruitment
was due to a paucity of eligible patients or excessively restrictive eligibility criteria.
Materials and Methods At 10 sites, eligible patients were retrospectively identified
by first screening blood bank databases over one month for all transfusion episodes
meeting trial inclusion criteria, defined as non-surgical patients receiving single RBC
unit transfusions. The age threshold was decreased from 65 to 50 years. The first 10
patients meeting inclusion criteria then underwent detailed chart review for the exclu-
sion criteria. The incidence of TACO and furosemide use was also recorded.
Results At the 10 sites, 11 969 red cell units were transfused over 1 month and 1356
met the inclusion criteria. Of the 100 charts reviewed, 60 (60%) had no exclusion cri-
teria. Active bleeding was the most common reason for ineligibility. There were 813
eligible transfusion episodes. Of the eligible patients, 17 (28
3%) had evidence of con-
gestive heart failure, and furosemide was prescribed in 24 (40%). Despite the use of a
lower age threshold, three cases of TACO were detected with an incidence of 3%.
Conclusion A large number of transfusion episodes met eligibility criteria. With a
3% incidence of TACO, 50% decrease through the use pre-transfusion furosemide
and a target consent rate of 30%, a definitive trial of approximately 3000
patients could be completed within 1 year.

Correspondence: Aditi Khandelwal MDCM FRCPC, Canadian Blood

Services, 67 College Street, Toronto, ON, Canada M5G 2M1
E-mail: aditi.khandelwal@blood.ca

1
2 A. Khandelwal et al.
Key words: furosemide, randomized clinical trials, transfusion-associated circula-
tory overload.
Introduction
Transfusion-associated circulatory overload (TACO) is the
leading cause of transfusion-related morbidity and mor-
tality. TACO is now the leading cause of transfusion-re-
lated morbidity [1,2] with up to 18% of patients requiring
transfer to intensive care and a 2% mortality rate [3]. Risk
factors for TACO appear to be largely intrinsic to the
recipient or the process of blood administration [3–5].
Patient risk factors include older age, pre-transfusion pos-
itive fluid balance, increased number of units transfused,
left ventricular systolic or diastolic dysfunction, history
of congestive heart failure and chronic kidney disease [3–
5]. As the recipient risk factors are often non-modifiable,
prophylactic strategies are limited to the avoidance of
unnecessary transfusion, the selection of lower-volume
blood product alternatives where feasible, slower infusion
rates and the co-administration of diuretics such as furo-
semide [4,6,7].
Although the pathophysiology of TACO is incompletely
understood, the hydrostatic pressure from the fluid vol-
ume transfused does not appear to be the sole considera-
tion [6,8]. TACO appears to be accompanied by signs of
inflammation in up to two-thirds of the individuals (e.g.
post-transfusion fever) which can further contribute to
endothelial injury [9,10]. The beneficial effect of loop
diuretics, such as furosemide, may also be more complex
than simply decreasing intravascular volume, as there is
evidence that intravenous furosemide can induce venodi-
lation [11,12]. Although pre-transfusion furosemide is
becoming increasingly popular as a means of preventing
TACO [13,14], there is little evidence to demonstrate its
effect on clinical outcomes, or guidelines on how to dose
appropriate so as to avoid harm from excessive diuresis.
As a result, peri-transfusion furosemide is not yet
endorsed by professional societies [15] and clinical equi-
poise remains as to whether pre-transfusion furosemide is
of benefit in preventing TACO.
To address this evidence gap, our group completed a
pilot study, TACO-BEL-1 (TACO-Best Eliminated with
Lasix), randomizing patients 65 years and older to pre-
transfusion furosemide versus placebo [16]. Feasibility
was not demonstrated at the two participating sites pri-
marily due to slower than anticipated enrolment. While a
low consent rate was partially responsible for this, the
primary limitation was the number of eligible patients
identified by screening. Whether this reflects excessively
restrictive eligibility criteria or a failure of the screening
protocol, however, was not clear. The present study was
therefore conducted to approximate the number of eligi-
ble patients, including an assessment of the effect of low-
ering the age cut-off from 65 to 50 years, by performing
a retrospective chart review, cross-referencing blood bank
and clinical patient records. Chart review also allowed for
a concurrent audit of patient risk factors for TACO, use of
peri-transfusion furosemide, and an assessment for non-
reported TACO cases.
Methods
Inclusion criteria for a trial of pre-transfusion furosemide
were defined as inpatients aged 50 years or older who
were transfused a single unit of RBCs within a 24-h per-
iod during regular working hours (weekdays 08:00 to
16:00). The reduced age cut-off of 50 years or older was
selected based on previous studies suggesting that the
incidence of TACO in this cohort is similar to those in
patients 65 or older [5]. Exclusion criteria were the same
as in the pilot trial [16] and included active bleeding (ei-
ther documented as such or with evidence of a haemoglo-
bin drop of greater than 20 mmol/l within 24 h), surgery
within 48 h, furosemide allergy, hemodynamic instability
(systolic blood pressure < 90 mmHg), hypokalaemia
(potassium < 3
0 mmol/l), hyponatraemia
(sodium < 130 mmol/l) and severe renal dysfunction (cre-
atinine clearance < 30 ml/min). The incidence of TACO in
this population was expected to be 3% [16]. A 50% risk
reduction in TACO with pre-transfusion furosemide was
hypothesized. Hence, 3066 total patients would be needed
to achieve a trial of 80% power to detect a statistically
significant effect with an alpha of 0
05.
Ten academic hospital sites across Canada were identi-
fied as potential sites for a larger RCT. Research ethics
board (REB) approval was obtained at all ten sites (British
Columbia 2, Ontario 7, Quebec 1) and informed consent
was waived by the REB. First, a 1-month audit of blood
transfusion service records was performed to identify
transfusion orders that met inclusion criteria during.
From these, the first 10 consecutive patients were selected
for detailed review for exclusion criteria, with the
assumption that the proportion of eligible patients in this
subgroup would be the same as the larger cohort of trans-
fusion orders that meeting inclusion criteria. The total
number of potentially enrolling patients at each site was
© 2020 International Society of Blood Transfusion
Vox Sanguinis (2020)
 TACO best eliminated with Lasix feasibility audit 3

determined by then multiplying this number by 20%,

which was the consent rate observed in the pilot trial 11969 RBC issued
at 10 sites
[16]. During the chart review process, evidence of TACO
[17] and risk factors for it (e.g. age, congestive heart fail- 4715 RBC issued for out-patients
ure, echocardiogram results and renal dysfunction) were 4069 RBC ordered after-hours
also collected. Finally, the co-administration of furose-
mide was also noted and recorded and converted into 3185 RBC issued to
inpatients between
approximate intravenous-equivalent dosing, by halving 8am and 4pm
the oral dose [18]. A descriptive analysis was performed
where normally distributed data were reported as mean 999 RBC issued to patients under
age 50 years
and standard deviations (SD), and skewed data were
reported as medians, interquartile ranges (IQR) and min/ 2186 RBC issued to
max using Microsoft Excel (Ver 16.33, 2019). Only de- patients with age
over 50 years
identified data were shared between the participating sites
and the coordinating centre. 830 RBC issued with other co-
components

Results
A total of 11 969 single RBC units were transfused dur-
ing a 1-month period at the 10 participating sites. As
shown in Fig. 1, 1356 (11
3%) of these orders met inclu-
sion criteria. All sites provided results for the retrospec-
tive chart review of 10 patients. Of the 100 patient
charts reviewed, 40 met one or more of the exclusion
criteria, most commonly active bleeding (20%) and sev-
ere renal dysfunction (12%), as shown in Fig. 2. Sixteen
of 40 (40%) had more than one exclusion criteria.
Demographic information for the 60 eligible patients is
presented in Table 1. The mean age of the eligible
patients was 71 – 10
2 years. Cardiac dysfunction was
documented in 17 (28%) and renal dysfunction in 15
(25%) of charts reviewed. Assuming that 60% of the
1356 RBC orders meeting inclusion criteria would simi-
larly not have exclusion criteria, it is estimated that the
10 participating sites could identify approximately 813
eligible patients per month. As the eligibility rate of
60% was determined by a convenience sample of 100
out of 1356 orders, the 95% confidence interval around
this measurement is 9
24%. Therefore, the number of eli-
gible patients can be estimated to range between 688
and 939. With a consent rate of 20%, this translates into
a range of 138 and 188 patients average 163) enrolled
per month. This estimate captures the collective enrol-
ment expectation of 10 hospital locations; differences in
patient demographics would presumably result in vari-
able enrolment rates between sites.
Furosemide was prescribed to 24 (40%) of patients
meeting eligibility criteria. Timing of furosemide could
only be determined in 20 of the 24 charts, with 12 orders
prescribed as pre-transfusion and 8 as post-transfusion.
The majority (20 of 24, 83%) of furosemide orders were
for intravenous administration (20 of 24, 83%). The med-
ian dose of 20 mg (range 10-120 mg).
1356 (11.3%)
RBC issued meet
study inclusion
criteria
Fig. 1 Assessment of RBCs issued over 1-month period at participating
sites for study inclusion criteria (RBC, red blood cell units).
100 individual
Reasons for exclusion:
charts reviewed
20 active bleeding
12 severe renal dysfunction
9 surgery within 48 h
6 hypotension
2 hypokalemia
60 individuals met
3 hyponatremia
inclusion criteria without
exclusion criteria
Fig. 2 Application of exclusion criteria to the 100 reviewed charts. Note
that some individuals had more than one exclusion criteria present.
Overall, three cases of TACO were identified amongst
the 100 charts reviewed, for an incidence of 3%. Two of
the three cases were not reported and/or not recognized
as a transfusion reaction before the chart review. Two of
the individuals had a history of cardiac dysfunction and
none of the individuals had renal dysfunction. In all
cases, furosemide was prescribed to treat respiratory dete-
rioration at variable doses (10, 40, 120 mg).
Discussion
In the previously published feasibility trial [16], the enrol-
ment of 80 patients at two participating sites over a 2-
month period was considered necessary for the successful
conduct of a large-scale definitive randomized controlled
trial, for which 10 sites would be expected to collectively
enrol 200 patients per month. The slower than anticipated
enrolment in that study (80 patients over 9 months) was
partially due to the low patient consent rate (anticipated

© 2020 International Society of Blood Transfusion

Vox Sanguinis (2020)
4 A. Khandelwal et al.
Parameter
Female sex (%)
Age (years) [median, IQR]
Weight (kg) [ median, IQR]
Cardiac dysfunction (%)a
Renal dysfunction (%)b
TACO [17] (%)
Furosemide use (%)
Furosemide dose in iv equivalent (mg) [median, IQR]
Abbreviations: TACO, transfusion associated circulatory overload.
a
Cardiac dysfunction was defined as a history of congestive heart failure documented in chart or
an echocardiogram report with left ventricular ejection fraction less than or equal to 50% or a
description of moderate to severe diastolic dysfunction.
b
Renal dysfunction was defined as a documented history of chronic renal dysfunction or dialysis
dependence or a creatinine clearance less than or equal to 30 ml/min. When a creatinine value
and weight was available, creatinine clearance was calculated using the Cockroft–Gault equa-
tion [21].
at 25%, observed at 20%), but primarily reflected the
unexpectedly low number of patients identified through
screening who met eligibility criteria. The current study
suggests that there are in fact a large number of eligible
transfusion episodes at the 10 participating sites: assum-
ing a 20% consent rate, 163 of the 814 eligible transfu-
sion episodes would be expected to complete the trial
protocol. While the inclusion criteria were broadened to
include patients 50 rather than 65 years or older, it is
notable most eligible patients identified in the current
study were still over age 65, which perhaps explains why
the incidence of TACO remained high at 3%. With an
enrolment rate of 163 patients per month at the 10 par-
ticipating hospitals, it would take approximately 1
5 years
to complete an adequately powered RCT of furosemide
for the prevention of TACO. If the consent rate could be
improved to 30%; however, the monthly enrolment rate
would increase to 244 and the trial could be completed in
approximately one year. Enrolment could be further
accelerated by expanding inclusion criteria to include
outpatients and extending screening to evenings and
weekends. However, this would increase the cost of the
study substantially. Alternatively, inclusion of patients
receiving multiple blood components would greatly
increase the number of eligible patients and improve its
overall clinical relevance. However, this in turn would
likely require the use of more variable furosemide dosing,
for which the development of a predictable dose-response
curve for this patient population would be necessary.
The screening protocol used in the pilot trial relied
upon receipt of a transfusion order by the hospital
blood bank to identify eligible patients. Often, the time
Table 1 Demographic information from chart
Eligible patients (N = 60) review presented for the eligible individuals
25 (42%)
71 [63–78]
81 [67
4–89
2]
17 (28%)
15 (25%)
3 (5%)
24 (40%)
40 [20–240]
interval between receipt of this order and the initiation
of the transfusion was too short to allow for patients
to be approached and consented for chart review for
exclusion criteria. This in turn appears to have greatly
underestimated the number of eligible patients. To
increase the time interval between identification of an
eligible patient and the initiation of the transfusion, an
alternative screening protocol may therefore be
required. One example would be to identify patients on
the basis of anaemia (e.g. haemoglobin < 90 g/l) rather
than an order to transfuse. This would not only allow
greater time to confirm patient eligibility and obtain consent
but would facilitate discussions with the attending physician
regarding the timing of any future transfusions so as to coor-
dinate the provision of pre-transfusion furosemide or match-
ing placebo.
An earlier audit conducted in 2010 revealed a much
lower rate of pre-transfusion furosemide with only 2
(0
6%) patients receiving pre-transfusion furosemide. Our
retrospective review showed that the use of furosemide is
as high as 40% amongst inpatients over the age of
50 years receiving 1 unit of red cell transfusion, with half
of the orders being prescribed pre-transfusion. During the
pilot TACO-BEL-1 study [16], one of the reasons for lower
consent rate was concerns regarding electrolyte imbal-
ances, hypovolaemia and acute renal failure due to furo-
semide. In the pilot protocol, prophylactic potassium
replacement was administered for patients with low
potassium to increase the physicians’ comfort level. No
significant electrolyte abnormalities, hypovolaemia or
renal dysfunction were seen in the pilot study [16] nor
the previously performed quality improvement initiative
© 2020 International Society of Blood Transfusion
Vox Sanguinis (2020)
 TACO best eliminated with Lasix feasibility audit 5

[13]. With the quick uptake of prophylactic furosemide
use, in any future trial, there is a risk for potential exclu-
sion of patients at high-risk of TACO. At present, there is
still evidence of clinical equipoise and we believe a
definitive trial is warranted to determine the value of this
common practice [19].
One of the challenges in determining accurate rates of
TACO remains the dependence of most haemovigilance
systems on passive reporting; for conditions such as
TACO, which can occur even in the absence of respiratory
distress [17] and which many clinicians might in fact not
consider a transfusion reaction per se, the result is both
under-recognition and under-reporting. With active
surveillance, by contrast, the incidence of TACO may be
as high as 5% [20]. Our retrospective chart review found
a rate of 3%, and 2 of the 3 cases had not been initially
recognized and/or reported as TACO. A secondary benefit
of undertaking a large trial in TACO might therefore be
raised awareness of TACO amongst hospital clinicians,
with improved reporting as a result. While this study was
not designed with the goal of identifying risk factors for
TACO, it is notable that a high proportion of patients who
met study inclusion criteria had cardiac and/or renal dys-
function.
In conclusion, with improved screening protocols and
improved consent rates, a definitive trial comparing pre-
transfusion furosemide to placebo for the prevention of
TACO could achieve target enrolment of approximately
3000 patients within 1 year. Clinical equipoise for furose-
mide remains. Our study shows that TACO, while under-
reported, remains a common transfusion hazard despite
widespread prophylactic use of furosemide.
Conflict of interests
The authors declare no conflict of interests.

References
1 Public Health Agency of Canada:
Health Canada Transfusion Transmit-
ted Injuries Surveillance System,
2006–2012 Report. Vol. 2016, 2014.
http://www.phac-aspc.gc.ca/hcai-ia
mss/ttiss-ssit/index-eng.php
2 FDA Center for Biologics Evaluation &
Research. Fatalities Reported to FDA
Following Blood Collection and Trans-
fusion: Annual Summary for Fiscal
Year 2017. 2020. Retrieved from
https://www.fda.gov/media/124796/d
ownload. February 9
3 Lieberma L, Maskens C, Cserti-Gazde-
wich C, et al.: A retrospective review
of patient factors, transfusion prac-
tices, and outcomes in patients with
transfusion-associated circulatory
overload. Transfus Med Rev 2013;
27:206–212
4 Alam A, Lin Y, Lima A, et al.: The pre-
vention of transfusion-associated cir-
culatory overload. Transfus Med Rev
2013; 27:105–112
5 Clifford L, Jia Q, Subramanian A,
et al.: Risk factors and clinical out-
comes associated with perioperative
transfusion-associated circulatory
overload. Anesthesiology 2017;
12:409–418
6 Andrzejewski C, Casey MA, Popovsky
MA: How we view and approach
transfusion-associated circulatory
overload: pathogenesis, diagnosis,
management, mitigation, and preven-
tion. Transfusion 2013; 53:3037–3047
7 Lin Y, Cohen R, Armali C, et al.:
Transfusion-associated circulatory
overload prevention: a retrospective
observational study of diuretic use.
Vox Sang 2018; 113:386–392
8 Bosboom JJ, Klanderman RB, Migdady
Y, et al.: Transfusion-associated car-
diac overload: a clinical perspective.
Transfus Med Rev 2019; 33:69–77
9 Parmar N, Pendergrast J, Lieberman L,
et al.: The association of fever with
transfusion-associated circulatory
overload. Vox Sang 2017; 112:70–78
10 Andrzejewski C Jr, Popovsky MA, Stec
TC, et al.: Hemotherapy bedside
biovigilance involving vital sign val-
ues and characteristics of patients with
suspected transfusion reactions associ-
ated with fluid challenges: can some
cases of transfusion associated circula-
tory overload have proinflammatory
aspects? Transfusion 2012; 52:2310–
2320
11 Jhund PS, McMurray JJ, Davie AP:
The acute vascular effects of furose-
mide in heart failure. Br J Clin Phar-
macol 2000; 50:9–13
12 ter Maaten JM, Dunning AM, Valente
MA, et al.: Diuretic response in acute
heart failure—an analysis from
ASCEND-HF. Am Heart J 2015;
170:313–321
13 Tseng E, Spradbrow J, Cao X, et al.:
An order set and checklist improve
physician transfusion ordering prac-
tices to mitigate the risk of transfu-
sion-associated circulatory overload.
Transfus Med 2016; 26:104–110
14 Fry JL, Arnold DM, Clase CM, et al.:
Transfusion premedication to prevent
acute transfusion reactions: a retro-
spective observational study to assess
current practices. Transfusion 2010;
50:1722–1730
15 AABB: Association Bulletin #15-02:
Transfusion Associated Circulatory
Overload. [monograph on the inter-
net]. 2015; Available from: http://
www.aabb.org/programs/publications/
bulletins/Docs/ab15-02.pdf [Last
accessed March 6, 2020]
16 Pendergrast J, Armali C, Cserti-Gazde-
wich C, et al.: Can furosemide prevent
transfusion-associated circulatory
overload? Results of a pilot, double-
blind, randomized controlled trial.
Transfusion 2019; 59:1997–2006
17 Popovsky M, Robillard P, Schipperus
M, et al.Proposed standard definitions
for surveillance of non infectious
adverse transfusion reactions. Interna-
tional Society of Blood Transfusion,
2011. http://www.isbtweb.org/filead
min/user_upload/Proposed_defini
tions_2011_surveillance_non_in
fectious_adverse_reactions_hae

© 2020 International Society of Blood Transfusion

Vox Sanguinis (2020)
6 A. Khandelwal et al.

movigilance_incl_TRALI_correction_
2013.pdf [Last accessed February 7,
2020]
18 Ponto LL, Schoenwald RD: Furosemide
(frusemide). A pharmacokinetic/phar-
macodynamic review (Part I). Clin
Pharmacokinet 1990; 18:381–408
19 Sarai M, Tejani AM: Loop diuretics for
patients receiving blood transfusions.
Cochrane Database Syst Rev 2015; 2:
CD010138
20 Roubinian NH, Hendrickson JE, Triulzi
DJ, et al.: Incidence and clinical char-
acteristics of transfusion-associated
circulatory overload using an active
surveillance algorithm. Vox Sang
2017; 112:56–63
21 Cockcroft DW, Gault MH: Prediction
of creatinine clearance from serum
creatinine. Nephron 1976; 16:
31–41

© 2020 International Society of Blood Transfusion

Vox Sanguinis (2020)