Journal of Critical Care 42 (2017) 138–146

Contents lists available at ScienceDirect

Journal of Critical Care

journal homepage: www.jccjournal.org

Renal

The effect of low-dose furosemide in critically ill patients with early

acute kidney injury: A pilot randomized blinded controlled trial
(the SPARK study)
Sean M. Bagshaw, MD, MSc a,⁎, R.T. Noel Gibney, MD a, Peter Kruger, MD, PhD b, Imran Hassan, MSc c,
Finlay A. McAlister, MD, MSc d, Rinaldo Bellomo, MD, PhD e
a
Department of Critical Care Medicine, Faculty of Medicine and Dentistry, University of Alberta, 2-124E Clinical Science Building, 8440-112 Street, Edmonton, Alberta T6G 2B7, Canada
b
Department of Intensive Care Medicine, Princess Alexandra Hospital, Ipswich Rd, Woolloongabba, and School of Medicine, University of Queensland, Queensland 4012, Australia
c
Epidemiology Coordinating and Research Centre (EPICORE), Department of Medicine, Faculty of Medicine and Dentistry, University of Alberta, 362 Heritage Medical Research Centre, Edmonton,
Alberta T6G 2S2, Canada
d
Division of General Internal Medicine, Faculty of Medicine and Dentistry, University of Alberta, 5-112 Clinical Sciences Building, 8440-112 Street, Edmonton, Alberta T6G 2B7, Canada
e
Department of Intensive Care, Austin Hospital, Studley Rd, Heidelberg, Victoria 3084, Australia

a r t i c l e i n f o a b s t r a c t

Keywords: Purpose: Furosemide is commonly prescribed in acute kidney injury (AKI). Prior studies have found conflicting
Acute kidney injury findings on whether furosemide modifies the course and outcome of AKI.
Furosemide Methods: Pilot multi-center randomized blinded placebo-controlled trial in adult patients with AKI admitted to
Urine output
three intensive care units. Participants were randomly allocated to furosemide bolus and infusion or 0.9% saline
Renal replacement therapy
Recovery
placebo. Primary endpoint was worsening AKI, defined by the RIFLE criteria. Secondary endpoints were kidney
Mortality recovery, renal replacement therapy (RRT) and adverse events.
Randomized Results: The trial was terminated after enrollment of 73 participants (37 to furosemide and 36 to placebo). Mean
Placebo (SD) age was 61.7 (14.3), 79.5% were medical admissions, mean (SD) APACHE II score was 26.6 (7.8), 90.4% re-
ceived mechanical ventilation and 61.6% received vasoactives. Groups were similar at baseline. No differences
were found in the proportion with worsening AKI (43.2% vs. 37.1%, p = 0.6), kidney recovery (29.7% vs. 42.9%,
p = 0.3), or RRT (27.0% s. 28.6%, p = 0.8). Adverse events, mostly electrolyte abnormalities, were more common
in furosemide-treated patients (p b 0.001). Protocol deviations were common, due often to supplementary furo-
semide.
Conclusions: In this pilot trial, furosemide did not reduce the rate of worsening AKI, improve recovery or reduce
RRT; however, was associated with greater electrolyte abnormalities.
Trial registration: ClinicalTrials.gov Identifier: NCT00978354 registered September 9, 2014.
© 2017 Elsevier Inc. All rights reserved.

1. Introduction
Acute kidney injury (AKI) is a frequent and increasingly encountered
complication of critical illness, with an incremental risk for morbidity
and chronic kidney disease (CKD) [1]. Despite investigation of numer-
ous preventative and therapeutic interventions to prevent or facilitate
⁎ Corresponding author.
E-mail addresses: bagshaw@ualberta.ca, @drseanbagshaw (S.M. Bagshaw),
ngibney@ualberta.ca (R.T.N. Gibney), Peter.Kruger@health.qld.gov.au (P. Kruger),
ihassan@ualberta.ca (I. Hassan), fmcalist@ualberta.ca (F.A. McAlister),
rinaldo.bellomo@austin.org.au (R. Bellomo).
repair and recovery for AKI, few, if any, have proven effective to favor-
ably modify the course or outcome once AKI is established [1-3]. How-
ever, there remain important knowledge gaps in our understanding of
the pathophysiology, case-mix identification, and response to interven-
tions commonly used in AKI [4].
Loop diuretics, most commonly furosemide [5-7], are received by an
estimated 60–70% of patients with AKI in intensive care unit (ICU) set-
tings, in spite of a contradictory evidence-base on their effectiveness
and safety [8-10].
Furosemide acts at the medullary thick ascending loop of Henle to
inhibit the Na+/K+/2Cl− pump on the luminal cell membrane surface.
Experimental data and small clinical studies have shown AKI increases
kidney metabolic demand and oxidative stress [11], and this may be at-
tenuated by furosemide [12-14]. Furosemide reduces renal tubular

http://dx.doi.org/10.1016/j.jcrc.2017.07.030
0883-9441/© 2017 Elsevier Inc. All rights reserved.
 S.M. Bagshaw et al. / Journal of Critical Care 42 (2017) 138–146
oxygen demand by impairing tubular Na+ reabsorption [14], and may
attenuate ischemia/reperfusion-induced apoptosis and associated gene
transcription in AKI [15,16]. In contrary, recent data have suggested
bolus furosemide aimed at augmenting urine output may increase kid-
ney oxidative stress [17].
Small clinical studies have further suggested that furosemide may
attenuate the severity or shorten the duration of AKI, and perhaps
delay or ameliorate need for renal replacement therapy (RRT); howev-
er, findings have been inconsistent and improvements in survival or
renal recovery have yet to be confirmed [8-10,18-22].
There has recently been recognition of the need for greater invest-
ment in strategies to improve outcomes in AKI from randomized trials
[4,23,24]. Arguably, scientific focus should also aim to improve our un-
derstanding of common therapies received by patients with AKI, such
as furosemide [17].
Accordingly, we performed a pilot randomized trial in critically ill
patients with mild AKI to compare the efficacy and safety of continuous
infusions of furosemide or placebo titrated to urine output on the inci-
dence of worsening AKI. We further aimed to address the feasibility of
such an approach, including gaining understanding of clinician equi-
poise for use of furosemide in critically ill patients with AKI.
2. Materials and methods
Details of the trial protocol were published and registered
(ClinicalTrials.gov Identifier: NCT00978354 registered September 14,
2009) [25]. The Human Research Ethics Committee at each site
reviewed and approved the study prior to commencement (University
of Alberta HREB file #Pro00002230). Written informed consent was ob-
tained from all participants or their legal representatives.
2.1. Study design, setting and participants
We performed a pilot randomized, blinded, placebo-controlled trial
at three tertiary university-affiliated hospital intensive care units
(ICU): General Systems Intensive Care Unit, University of Alberta Hospi-
tal (Edmonton, Canada), Intensive Care Unit, Austin Hospital (Mel-
bourne, Australia) and Intensive Care Unit, Princess Alexandra
Hospital (Brisbane, Australia).
We screened patients admitted to a study ICU for eligibility. Inclusion
Criteria (all patients were required to fulfill the following criteria): 1)
evidence of early AKI (RIFLE category – RISK); 2) peripheral or central
intravenous catheter and urinary catheter; 3) ≥2 criteria for the system-
ic inflammatory response syndrome (SIRS) within 24 h of screening;
and 4) achieved immediate resuscitation goals, based on the judgment
of the most responsible treating physician and including one or more
of the following: fluid resuscitation and/or vasoactive therapy to
achieve mean arterial pressure (MAP) ≥65 mm Hg, central venous pres-
sure (CVP) ≥8 cm H2O, central venous oxygen saturation (ScvO2) ≥70%
(if measured) and/or cardiac index (CI) ≥ 2.5 L/min/1.73 m2 (if mea-
sured). Exclusion criteria included any of: 1) age b 18 years; 2) con-
firmed or suspected pregnancy (verified by serum [β-HCG] pregnancy
test if indicated); 3) suspected or confirmed obstructive etiology for
AKI; 4) ≥stage 4 chronic kidney disease (CKD) (defined by an estimated
glomerular filtration rate (eGFR) b30 mL/min/1.73 m2), end stage kid-
ney disease (ESKD) receiving maintenance dialysis or kidney transplan-
tation; 5) recent RRT during ICU or index hospitalization; 6) recovering
AKI, defined as a ≥ 25% or ≥ 44.2 μmol/L decline from peak increase in
serum creatinine; 7) acute pulmonary edema mandating urgent furose-
mide administration or RRT initiation or patient was already receiving a
continuous furosemide infusion; 8) moribund status with expected
death within 24 h or significant limitations of medical therapy
(LOMT); 9) suspected or known allergy to furosemide; and 10) prior
enrolment.
139
2.2. Randomization and blinding
Eligible participants were randomly allocated in a 1:1 ratio to furose-
mide (bolus followed by continuous infusion) or placebo-control (0.9%
saline). Computerized random blocks of variable size (4–8) were strat-
ified within each site and by the presence of sepsis. Randomization was
performed by an unblinded clinical trial pharmacist using a web-based
application through the Epidemiology Coordinating and Research Cen-
tre (EPICORE) at the University of Alberta (available at: https://www.
epicore.ualberta.ca/home/). This was concealed from study
investigators.
2.3. Procedures
The furosemide administration protocol was adapted from a prior
trial [26]. Each furosemide infusion bag contained 2000 mg of furose-
mide in 500 mL of 0.9% saline for a final concentration of 4 mg/mL
(Item S1). To ensure blinding of investigators and treating clinicians,
study infusion bags for furosemide and placebo appeared identical and
were identified by a 4-digit coded identifier.
Each participant received a weight-based loading dose of 0.4 mg/kg
followed by a continuous infusion commenced at a dose of
0.05 mg/kg/h. Estimated ideal body weight (IBW), based on the formula
described by Devine [27], was used to determine the urine output target
(Tables S2–S4). The study infusion was titrated to achieve and maintain
a target urine output in the range of 1.0–2.0 mL/kg/h. The maximum in-
fusion rate was 0.40 mg/kg/h. Urine output was assessed hourly. The
treatment algorithm for titration of the continuous infusion is shown
in the Table S5.
The study infusion was continued for a minimum of 24 h and
discontinued if any one of the following events occurred: 1) kidney re-
covery; 2) decision to start RRT; 3) death; 4) adverse reaction attributed
to study intervention; 5) ICU discharge; or 6) a total of 7-days of study
drug had been administered.
All other aspects of patient management within the parameters
outlined (i.e., methods of fluid resuscitation, choice of fluids, vasoactive
therapy, choice of vasoactive therapy) were left to the discretion of the
treating ICU team. No additional interventions were performed.
2.4. Operational definitions
Acute kidney injury (AKI) – AKI was defined and classified according
to the RIFLE criteria (acronym indicating Risk of renal dysfunction; Inju-
ry to the kidney; Failure of kidney function; Loss of kidney function; and
End-stage kidney disease) [28]. The presence of early AKI was defined by
a minimum of RIFLE category – RISK. Renal replacement therapy (RRT) –
RRT was defined as any form of extracorporeal kidney support provided
to patients with documented AKI. By protocol and in order to minimize
the potential bias of clinician discretion on when to initiate RRT, at least
one conventional indications for RRT was required to be fulfilled prior to
starting RRT (Item S1). Renal recovery – For the trial, renal recovery was
defined as the return of serum creatinine to within 10% of baseline levels
and a spontaneous urine output ≥1.0 mL/kg/h for a minimum of 24 h
and not receiving RRT. Systemic inflammatory response syndrome (SIRS)
– To fulfill SIRS, any 2 or more of the SIRS criteria were required to be
fulfilled (see Item S1) [29]. Sepsis – The clinical syndrome of sepsis
was defined by the presence of suspected or confirmed infection and
the presence of ≥2 SIRS criteria.
2.5. Endpoints
The primary endpoint was worsening AKI, defined as progression
from RIFLE category – RISK to a more severe category of AKI (i.e.,
RIFLE categories – INJURY or FAILURE or receipt of RRT) in the 7 days fol-
lowing randomization. Pre-specified secondary endpoints compared
differences in cumulative fluid balance, serum electrolytes (i.e.,
140 S.M. Bagshaw et al. / Journal of Critical Care 42 (2017) 138–146
potassium, magnesium), acid-base status (i.e., pH, serum bicarbonate),
rates of RRT initiation, rates of renal recovery, and hospital mortality be-
tween furosemide and placebo groups, respectively.
2.6. Statistical analysis
Based on observational data, we conservatively estimated that 60%
of critically ill patients with AKI would worsen and progress [30]. We es-
timated that furosemide would contribute to a 20% absolute reduction
in the proportion who progress from RIFLE class – RISK. This required
a sample size of 214 patients to provide 80% power (alpha 0.05) [25].
Statistical analysis was performed using the principle of intention-
to-treat. Normally or near normally distributed variables are reported
as means with standard deviations (SD) and compared using the t-
test. Continuous data that were far from normality were reported as
medians with inter-quartile ranges (IQR) and compared using the
Mann Whitney U test. Hodge-Lehmann estimate of location shift with
its confidence interval was reported along with the Mann Whitney U
test's p-value. Categorical data, including the primary endpoint, and
need for RRT, renal recovery and hospital mortality, were reported as
proportions and compared using Chi-square test. Fisher's Exact test
was used when small expected frequencies invalidates the Chi-square
test. Multi-variable logistic regression was used to control for confound-
ing due to imbalances in baseline characteristics after randomization.
2.7. Role of the funding source
The funder had no role in the design, data collection, data analysis,
interpretation or in the writing of the manuscript. The corresponding
author (SMB) and biostatistician (IH) had full access to the data.
3. Results
Between September 1, 2009 and June 2, 2014, we enrolled 73 partic-
ipants from 3 sites: 37 were randomly allocated to receive furosemide
and 36 to placebo (Fig. 1).
The trial was terminated early due to feasibility of achieving recruit-
ment following two unavoidable enrollment disruptions to trial
Fig. 1. Trial profile (consort diagram).
recruitment and funding limitations. The termination was supported
by the Data Safety Monitoring Committee. These disruptions included:
1) secondment of research personnel to bedside clinical duties during
pandemic influenza A(H1N1) in October 2009 (6 months); and 2) a
North American shortage of furosemide due to generic drug maker
Sandoz Canada suspending production in October 2011, along with in-
ability to import furosemide and restricted access imposed by Alberta
Health Services (12 months).
No patients withdrew consent. One patient allocated to the placebo
arm did not receive the study intervention. The primary endpoint was
available for all 72 participants and included in the final intention-to-
treat analysis.
3.1. Baseline characteristics
The mean (SD) age of enrolled participants was 61.7 (14.3), 16
(21.9%) were female, mean (SD) APACHE II score was 26.6 (7.8), mean
(SD) sequential organ failure assessment (SOFA) score was 10.8 (3.7),
58 (79.5%) were medical admissions, 66 (90.4%) were receiving me-
chanical ventilation and 45 (61.6%) were receiving vasoactive therapy,
respectively.
Following randomization, baseline characteristics, acute physiology,
laboratory parameters and treatment intensity were comparable be-
tween furosemide and placebo groups (Tables 1 and 2). Participants al-
located to furosemide had more exposure to vancomycin (67.6% vs.
38.9%, p = 0.01).
3.2. Processes of care
In the 72 h prior to randomization, 30 patients (41.0%) had been ex-
posed to furosemide (38.9% for furosemide vs. 48.5% for placebo, p =
0.5). Urine output in the 6 h preceding randomization was similar be-
tween the furosemide (210 mL [160–505]) and placebo (235 mL
[154–443]) groups (p = 0.9). In total, 69.4% had an episode of oliguria
(≤0.5 mL/kg/h × 6 h) in the 72 h prior to randomization (77.8% for furo-
semide vs. 61.1% for placebo, p = 0.3). Cumulative fluid balance in the
24 h prior to randomization was a median (IQR) 2741 mL (1136–
 S.M. Bagshaw et al. / Journal of Critical Care 42 (2017) 138–146 141

Table 1 Table 2
Baseline clinical characteristics by treatment arm. Pre-randomization baseline physiological and laboratory values by treatment arm.

Variable Furosemide Placebo Variable Furosemide Placebo

(n = 37) (n = 36) (n = 37) (n = 36)

Age (years), (median, IQR) 61.9 (49.5–69.6) 65.5 (54.9–71.2) Serum creatinine (mg/dL) (median, IQR) 1.69 (1.07–2.79) 1.88 (1.13–2.63)
Male sex (n, %) 30 (81.1) 27 (75.0) (n = 37) (n = 36)
Weight (kg), (median, IQR) 82.0 (70.0–100.0) 83.7 (72.3–95.3) Urine output (mL) 210 (160–505) 235 (154–443)
No. of comorbid diseases (median, IQR) 3 (2–4) 3.5 (2–5) (6 h pre-randomization) (median, IQR) (n = 37) (n = 36)
Susceptibilities for AKI Fluid balance (6 h pre-randomization) 709 (434–954) 505 (248–988)
Diabetes mellitus (n, %) 7 (19.4) 10 (28.6) (median, IQR) (n = 37) (n = 35)
Chronic liver disease (n, %) 6 (16.7) 5 (14.3) Cumulative fluid balance 6363 (3325–9496) 5906 (2760–8979)
Heart failure (n, %) 4 (11.1) 2 (5.7) (from ICU admission) (median, IQR) (n = 37) (n = 35)
Chronic kidney disease (n, %) 0 (0) 2 (5.7) Blood urea nitrogen (mg/dL) 34.2 (17.9–52.4) 34.2 (25.4–45.7)
Hospital exposures for AKI (median, IQR) (n = 36) (n = 36)
ACEi/ARB (n, %) 10 (27.8) 11 (31.4) Serum Sodium (mmol/L) (median, IQR) 141 (136–143) 139 (134–142)
NSAID (n, %) 5 (13.5) 10 (27.8) (n = 36) (n = 36)
Radiocontrast media (n, %) 12 (33.3) 12 (34.3) Serum Potassium (mmol/L) 3.8 (3.6–4.4) 4.1 (3.6–4.8)
Aminoglycoside/amphotericin (n, %) 4 (11.1) 4 (11.1) (median, IQR) (n = 36) (n = 36)
Vancomycin (n, %) 25 (67.6) 4 (11.1) Serum Chloride (mmol/L) (median, IQR) 110 (102−113) 107 (103−111)
Rhabdomyolysis (CK N 5000 U/L) (n, %) 0 (0) 2 (5.6) (n = 36) (n = 36)
Primary diagnostic category Serum Bicarbonate (mEq/L) 22.7 (20.3–24.1) 22.8 (19.9–25.1)
Cardiovascular (n, %) 4 (10.8) 4 (11.1) (median, IQR) (n = 36) (n = 36)
Respiratory (n, %) 10 (27.0) 13 (36.1) Serum Magnesium (mmol/L) 0.99 (0.81–1.15) 1.01 (0.82–1.17)
Gastrointestinal (n, %) 1 (2.7) 5 (13.9) (median, IQR) (n = 32) (n = 35)
Neurological (n, %) 6 (16.2) 1 (2.8) Serum pH (median, IQR) 7.40 (7.36–7.44) 7.40 (7.34–7.46)
Sepsis (n, %) 12 (32.4) 6 (16.7) (n = 35) (n = 35)
Trauma (n, %) 3 (8.1) 3 (8.3) Base Excess (median, IQR) −1.0 (−4.0–1.0) −1.4 (−6.0–2.0)
Metabolic/poisoning (n, %) 0 3 (8.3) (n = 35) (n = 34)
Hematologic (n, %) 1 (2.7) 0 Serum hemoglobin (lowest) (g/L) 96.0 (90.0–111.0) 105.0 (89.0–120.0)
Other (n, %) 0 1 (2.8) (median, IQR) (n = 36) (n = 36)
Admission type Serum platelets (lowest) (109/mL) 132.0 (73.0–215.0) 146.0 (88.0–227.0)
Medical (n, %) 31 (83.8) 27 (75.0) (median, IQR) (n = 35) (n = 35)
Emergency surgical (n, %) 5 (13.5) 4 (11.1) White cell count (highest) (109/mL) 12.5 (7.7–17.0) 12.8 (8.5–18.0)
Elective surgical (n, %) 1 (2.7) 5 (13.9) (median, IQR) (n = 36) (n = 36)
APACHE II score (mean, SD) 25.6 (7.4) 27.7 (8.1) Serum bilirubin (highest) (mg/dL) 1.55 (0.82–2.92) 1.32 (0.76–2.22)
SOFA score (mean, SD) 10.9 (3.5) 10.7 (4.0) (median, IQR) (n = 34) (n = 36)
Mechanical ventilation (n, %) 33 (89.2) 33 (91.7) Serum lactate (highest) (mmol/L) 2.5 (1.4–3.8) 2.60 (1.65–4.60)
Vasopressors/inotropes (n, %) 23 (62.2) 22 (62.9) (median, IQR) (n = 35) (n = 32)
FiO2 (highest) (%) (median, IQR) 0.65 (0.45–1.00) 0.70 (0.60–0.80)
Abbreviations: IQR = intraquartile range; AKI = acute kidney injury; ACEi = ACE inhibi-
(n = 34) (n = 33)
tor; ARB = angiotensinogen receptor blocker; NSAID = non-steroidal anti-inflammatory
PaO2 (lowest) (mm Hg) (median, IQR) 75.0 (65.0–86.0) 71.0 (57.0–81.0)
drug; APACHE = Acute Physiology and Chronic Health Evaluation; SOFA = Sequential
(n = 35) (n = 34)
Organ Failure Assessment.
PaCO2 (highest) (mm Hg) (median, IQR) 44.0 (37.0–49.0) 44.0 (39.0–49.0)
(n = 34) (n = 34)

4512) and 2002 mL (755–4479) in the furosemide and placebo groups

(p = 0.5), respectively.
The median (IQR) duration of study treatment was 2 days (2–3) in
the furosemide group and 3 days (2–5) in the placebo group (p =
0.2). Among those allocated to furosemide, the median (IQR) total
dose of furosemide was 1.65 mg/kg (0.7–4.8) and 3.2 mg/kg (1.4–
16.1) in the first 24 h and over the study period, respectively. The
urine output in the first 6 h (745 mL [369–1360] vs. 210 mL [149–
425], p b 0.001) and first 24 h (3006 mL [1475–5035] vs. 1165 [758–
2100], p = 0.002) after randomization was significantly greater in the
furosemide compared with placebo group.
There were no significant differences in organ failure scores, hemo-
dynamic profile, or fluid balance between groups during the study inter-
vention (Fig. 2). While there were no significant differences in serum
creatinine and potassium levels during the study intervention, those al-
located to furosemide showed higher serum sodium and bicarbonate
levels (Fig. 3).
Similar proportions of patients in furosemide and placebo groups
received potassium supplementation in the first 24 h (46.0% vs. 48.6%,
p = 1.0) and 48 h (62.5% vs. 60.0%, p = 1.0) after randomization; how-
ever, the furosemide group received greater a cumulative dose over the
first 48 h (124 mmol [60–206] vs. 80 mmol [31 − 110], p = 0.006).
There was no difference in the proportion receiving magnesium
supplementation (24 h: 18.9% vs. 17.1%, p = 1.0; 48 h: 12.5% vs.
20.0%, p = 0.5) or in the cumulative dose in the first 48 h (5 g [3-5]
vs. 5 g [2-8], p = 0.5).
3.3. Primary and secondary endpoints
There was no statistically significant difference in the primary end-
point of worsening AKI between furosemide and placebo groups (16
[43.2%] vs. 13 [37.1%]; OR 1.52, 95% CI, 0.61–3.83, p = 0.48) (Table 3).
There was also no difference in worsening AKI in the subgroup with sep-
sis (14 [43.8%] vs. 11 [37.9%]; OR 1.28; 95% CI 0.50–3.30). In multivari-
able analysis, adjusting for APACHE II score (OR 1.14; 95% CI, 0.42–
3.00), vancomycin exposure (OR 1.08; 0.39–2.98), pre-randomization
exposure to furosemide (OR 1.39; 95% CI, 0.52–3.76) or pre-randomiza-
tion oliguria (OR 1.18; 95% CI, 0.45–3.15, p = 0.7) made no difference.
No statistically significant differences were observed across second-
ary endpoints, including kidney recovery, RRT utilization, mortality or a
composite of RRT and/or mortality (Table 3). While there was potential-
ly a clinically meaningful difference in cumulative fluid balance
(−1081 mL [95% CI, −2697 to 467]), this was not statistically signifi-
cant. There were no significant between group differences in maximal
changes in electrolytes and acid-base status (Table 4).
3.4. Adverse events and protocol deviations
There was no difference in the proportion of patients with adverse
events (p = 0.6); however, the incidence rate was higher in those re-
ceiving furosemide (p b 0.001) (Table 5). In the furosemide arm, 25
142 S.M. Bagshaw et al. / Journal of Critical Care 42 (2017) 138–146
Fig. 2. Changes in daily organ failure, hemodynamic parameters and fluid balance stratified by study intervention. A) SOFA score; B) mean arterial pressure; C) central venous pressure; and
D) daily fluid balance (Error bars show 95% CI for mean differences between groups on each day of study intervention).
(67.6%) participants had 111 adverse events reported (incidence rate
4.4 events/patient), of which 2 (5.4%) contributed to study intervention
discontinuation. In the placebo arm, 22 (61.1%) participants had 57 ad-
verse events reported (incidence rate 2.6 events/patient), of which 1
(2.8%) contributed to study intervention discontinuation. The majority
of adverse events were related to occurrence of electrolyte abnormali-
ties in both groups. Protocol deviations were common; however, not
significantly different between groups (p = 0.8). The majority were re-
lated to deviation in the treatment algorithm. Compared with the furo-
semide arm, those allocated to placebo were more likely to receive
supplementary diuretic therapy (10.8% vs. 30.6% vs. p = 0.046).
4. Discussion
There is a pathophysiological mechanistic basis for giving low-dose
furosemide in early AKI to reduce kidney metabolic work [14-16] and
we hypothesized that it may mitigate worsening AKI and facilitate ear-
lier recovery of kidney function. We tested this hypothesis in a multi-
center pilot randomized trial of critically ill patients with early AKI com-
paring continuous furosemide infusion to placebo-control on the occur-
rence of worsening AKI [25].
We found worsening AKI was common in this cohort of critically ill
patients with AKI, occurring in 39.7%, but with no difference between
the furosemide and placebo-control groups (43.2% vs. 37.1%, p = 0.6),
respectively. We did not find any clinically important differences in
key secondary endpoints including kidney recovery, duration of AKI,
or mortality. RRT use was also frequent, initiated in 27.4% of patients;
however, we found no differences between furosemide and placebo-
treated groups. A substantial number of adverse events occurred, pre-
dominantly attributable to minor electrolyte imbalances, with rates
higher among those receiving furosemide. Similarly, protocol deviations
frequently occurred in both groups. These were largely due to non-ad-
herence to the study algorithm and extra (i.e., non-protocol) furose-
mide use.
4.1. Context with prior literature
The necessity for additional clinical trials specifically focused on
unraveling the optimal use, efficacy and safety of furosemide in AKI
stems from an apparent disconnect in the widespread utilization of fu-
rosemide in practice juxtaposed with evidence suggestive of no clinical
benefit and potential harm [5,31-33]. Early clinical trials of furosemide
use in AKI were small, confounded by co-interventions (i.e., mannitol,
dopamine), characterized by late administration (i.e., long durations of
oliguria or already receiving RRT), and by use of large bolus furosemide
doses with no specific titration of therapy to physiologic endpoints such
as urine output [8]. Such data lack generalizability to AKI and current fu-
rosemide practices in ICU settings.
Observational studies detailing the risk of furosemide for develop-
ment of AKI or poor outcome associated with its use are almost certainly
susceptible to confounding by indication [5,33]. In these studies, the risk
of death or non-recovery was predominantly borne by patients with
more severe oliguric AKI who did not show a robust response to initial
diuretic challenge [5,33]. It is likely many of these patients had delay
 S.M. Bagshaw et al. / Journal of Critical Care 42 (2017) 138–146 143

Fig. 3. Changes to daily biochemical profile stratified by study intervention. A) in serum creatinine; B) serum sodium; C) serum potassium; and D) serum bicarbonate (Error bars show 95%
CI for mean differences between groups on each day of study intervention).

to timely start of RRT or had unmeasured differences in probability of Indeed, in a cohort of acute lung injured patients with concomitant
being offered RRT. Consistent with the findings of our pilot trial, addi- AKI, after adjusting for fluid balance, there was no ceiling dose of furose-
tional observational studies have not shown significant differences in mide above which mortality was shown to increase [34].
outcomes associated with furosemide exposure and have failed to con- While our pilot trial was clearly underpowered and we did not show
firm any specific incremental risk with furosemide use in AKI [6,7,34]. furosemide reduced the severity or shortened the duration of AKI, furo-
semide may still have an important role for managing fluid balance and
to counterbalance the obligatory fluid intake and optimal delivery of nu-
trition routinely given to critically ill AKI patients [35,36]. Moreover,
Table 3
Summary of study outcomes.
loop diuretics such as furosemide are a common strategy used by clini-
cians to avert starting RRT among those with severe AKI [37,38]. How-
Variable Furosemide Placebo OR/difference p-Value ever, it remains uncertain whether furosemide exposure in this
(n = 37) (n = 36) (95% CI)
context portends risk or benefit for kidney recovery or mortality.
Worsened AKI (n, %)a 16 (43.2) 13 (37.1) – 0.6
Cumulative fluid balance 877 (−515 2407 (−522 −1081 0.2
(mL) (med [IQR]) to 2920) to 4383) (−2697 to
467) 4.2. Implications for clinicians and research
Kidney recovery (n, %)b 11 (29.7) 15 (42.9) – 0.3
Time to recovery (h) 48.6 54.6 4 (−33.5 to 0.7 Despite survey data suggesting clinicians had sufficient equipoise to
(median, IQR) (42.9–82.4) (43.5–82.7) 19.4) enroll AKI patients in such a trial, we found this protocol was challeng-
Received RRT (n, %) 10 (27.0) 10 (28.6) – 0.9
ing to implement [39,40]. Moreover, our data suggested a continuous
ICU death (n, %) 3 (8.1) 6 (17.1) – 0.3
Hospital death (n, %) 3 (8.1) 5 (14.3) – 0.5 furosemide infusion is not likely to prevent worsening AKI; however,
Mortality at 90-days (n, 8 (21.6) 11 (30.5) – 0.4 it may still represent an important approach to maintaining fluid ho-
%) meostasis. Clinicians appeared concerned about patient fluid status with
Received RRT or death at 17 (46.0) 19 (54.3) – 0.5
numerous occurrences of non-protocol administration of furosemide
90-days (n, %)
[35]. We believe these observations align with recent data showing di-
a
Worsening RIFLE category (primary endpoint) - defined as any worsening in RIFLE uretic therapy did not confer incremental risk of death or non-recovery
category and/or receipt of RRT.
b
Renal recovery – the operational definition of renal recovery will be the return of se-
in selected critically ill patients when aimed at mitigating fluid accumula-
rum creatinine to within 10% of baseline levels and a spontaneous urine output tion and complications of fluid overload [6,34]. Furosemide use in ICU set-
≥1.0 mL/kg/h for a minimum of 24 h independent of RRT. tings is predominantly triggered in response to fluid accumulation in the
144 S.M. Bagshaw et al. / Journal of Critical Care 42 (2017) 138–146

Table 4
Summary of physiological/laboratory process outcomes.

Variablea Furosemide Placebo Difference p-Value

(n = 37) (n = 36) (95% CI)

Largest Δ serum potassium (mean, SD) 0.59 (0.68) 0.62 (0.65) −0.03 (−0.34 to 0.29) 0.9
(n = 36) (n = 35)
Largest Δ serum magnesium (mean, SD) 0.13 (0.20) 0.07 (0.28) 0.06 (−0.06 to 0.18) 0.3
(n = 32) (n = 34)
Largest Δ serum pH (mean, SD) 0.03 (0.06) 0.06 (0.07) −0.02 (−0.06 to 0.01) 0.2
(n = 35) (n = 33)
Largest Δ serum bicarbonate (mean, SD) 0.35 (3.29) 0.30 (2.90) 0.05 (−1.42 to 1.52) 0.9
(n = 36) (n = 35)
a
Largest Δ referenced from baseline value at randomization to lowest observed value.

context of secondary organ injury (i.e., acute lung injury) rather than iso-
lated AKI per se [40]. More recently, the urine output response to a bolus
dose of furosemide (i.e., furosemide stress test [FST]) has shown utility as
a functional dynamic measure to predict AKI severity [41,42]. In critically
ill patients with AKI, non-response to a furosemide challenge may repre-
sent a logical prompt for support escalation and starting RRT [36,43]. Fu-
ture research should focus on the ideal interventional strategies for fluid
balance management incorporating the use loop diuretics among critical-
ly ill patients with AKI [36].
4.3. Strengths and limitations
Our study had a number of strengths, including being multi-centric,
randomized, placebo-controlled, obtaining complete ascertainment of
primary endpoints and evaluating a biologically plausible and titratable
approach to furosemide administration in critically ill patients with AKI
[25]. However, our trial also has noteworthy limitations. Our trial was
small and was prematurely terminated, and as such, is predisposed to
type I error and has suboptimal statistical power. In addition, we expe-
rienced a high rate of protocol deviations, highlighting the challenges
with a trial of furosemide administration in critically ill patients. Finally,
while allocation was concealed and the intervention blinded, it is plau-
sible that unblinding may have occurred in selected patients with brisk
urine output responses to receiving furosemide; however, our primary
outcome was objectively measured and analysts were blinded to
group assignment.
5. Conclusions
In summary, in this pilot study, furosemide infusion in critically ill
patients with early AKI did not reduce the incidence of worsening AKI
or favorably impact any secondary endpoints including kidney recovery,
duration of AKI, cumulative fluid balance, RRT use or mortality. In the
ICU setting, further research should aim to evaluate the ideal contribu-
tion for furosemide in the routine management of critically ill patients
with AKI, such as maintaining fluid balance homeostasis.

Table 5
Summary of adverse events and protocol deviations/violations.

Variable Furosemide Placebo

(n = 37) (n = 36)

No. events No. patients No. patients discontinued No. events No. patients No. patients discontinued
(n = 111) (n = 25) (n = 2) (n = 57) (n = 22) (n = 1)

Adverse eventsa
Drug reaction 1 1 1 1 1 1
Tinnitus 0 0 0 1 1 0
Arrhythmia – SVT 4 3 0 5 4 0
Arrhythmia – VT/VF 2 2 0 0 0 0
Cardiopulmonary arrest 0 0 0 0 0 0
Other 0 0 0 1 2 1
Electrolyte abnormalities 104b 22c 1 49b 14c 0
Serum sodium ≥ 150 mmol/L 53 9 1 1 1 0
Serum potassium b 3.0 mmol/L 7 4 0 2 2 0
Serum magnesium b 0.7 mmol/L 2 2 0 6 4 0
Serum bicarbonate ≥ 30 or pH ≥ 7.5 62 18 1 40 10 0

Variable Furosemide Placebo

(n = 37) (n = 36)

No. events No. patients No. patients discontinued No. events No. patients No. patients discontinued
(n = 171) (n = 28) (n = 7) (n = 201) (n = 29) (n = 3)

Protocol deviation/violationa
Did not receive study intervention 0 0 0 1 1 1
Supplementary diuretic therapy given 12 4 0 76 11 0
Study intervention interrupted 4 4 0 7 6 1
Study intervention discontinued 7 7 7 2 2 0
Study algorithm deviation 123 22 0 103 18 1
Other 25 11 0 12 6 0
a
Numbers in columns of adverse events and protocol deviations/violations are not mutually exclusive.
b
More than one electrolyte abnormality was reported with each discrete adverse event episode.
c
More than one adverse event was reported per patient.
 S.M. Bagshaw et al. / Journal of Critical Care 42 (2017) 138–146 145

Abbreviations Nephrology. FAM was supported by a Senior Health Scholar Award from
AKI acute kidney injury Alberta Innovates – Health Solutions and the Capital Health/University
CKD chronic kidney disease of Alberta Chair in Cardiovascular Outcomes Research. We thank Dr.
CVP central venous pressure Ross Tsuyuki (Director) and the staff at EPICORE for facilitation data co-
ESKD end-stage kidney disease ordination and management. We thank Dr. Andrew Shaw (chair), Dr.
IBW ideal body weight Patrick Murray and Dr. Paul Miles for serving on the Data Safety Moni-
ICU intensive care unit toring Committee.
IQR intraquartile range The following are the participating centers, investigators and coordi-
LOMT limitation of medical therapy nators for the SPARK study: University of Alberta Hospital (Edmonton):
MAP mean arterial pressure Sean M. Bagshaw, R.T. Noel Gibney, Finlay A. McAlister, Samantha Tay-
RIFLE risk, injury, failure, loss, end-stage lor, Nadia Baig; Kristen Reid; Austin Hospital (Melbourne): Rinaldo
RRT renal replacement therapy Bellomo, Glenn Eastwood; Princess Alexandra Hospital (Brisbane): Peter
ScvO2 central venous oxygen saturation Kruger; Jason Meyer.
SD standard deviation
SIRS systemic inflammatory response syndrome Appendix A. Supplementary data
SOFA sequential organ failure assessment
Supplementary data to this article can be found online at http://dx.
Ethics approval and consent to participate doi.org/10.1016/j.jcrc.2017.07.030.

The Human Research Ethics Committee at each site reviewed and
approved the study prior to commencement (University of Alberta
HREB file # Pro00002230). Written informed consent was obtained
from all participants or their legal representatives.
Consent for publication
Not applicable.
Availability of data and material
The datasets used and/or analyzed during the current study are
available from the corresponding author on reasonable request.
Competing interests
SMB reports having received honoraria and has consulted for Baxter
Healthcare Corp.
Funding
This work was supported by a grant from the Alberta Innovates –
Health Solutions (Grant File No. 200700566) and Intensivist Group in
the Department of Critical Care Medicine, Faculty of Medicine and Den-
tistry, University of Alberta. The funding agency had no role in the de-
sign or conduct of the study, in the collection, management, analysis
or interpretation of the data, or in the preparation, review or approval
of the manuscript.
Authors' contributions
SMB designed the trial, obtained funding for the trial, obtained data,
analyzed the data, interpreted data, drafted the manuscript and provid-
ed critical revision of the manuscript. FAM designed the trial, obtained
funding for the trial, interpreted data, and provided critical revision of
the manuscript. RB designed the trial, obtained data, obtained funding
for the trial, interpreted data, and provided critical revision of the man-
uscript. RTNG obtained funding for the trial, interpreted data, obtained
data, and provided critical revision of the manuscript. IH analyzed and
interpreted data, and provided critical revision of the manuscript. PK
interpreted data, and provided critical revision of the manuscript. All
authors have ready and approved the final manuscript.
Acknowledgements
SMB was supported by a Clinical Investigator Award from Alberta In-
novates – Health Solutions and a Canada Research Chair in Critical Care
References
[1] Rewa O, Bagshaw SM. Acute kidney injury-epidemiology, outcomes and economics.
Nat Rev Nephrol 2014;10(4):193–207.
[2] Abel RM, Beck Jr CH, Abbott WM, Ryan Jr JA, Barnett GO, Fischer JE. Improved surviv-
al from acute renal failure after treatment with intravenous essential L-amino acids
and glucose. Results of a prospective, double-blind study. N Engl J Med 1973;
288(14):695–9.
[3] Lameire NH, Bagga A, Cruz D, De Maeseneer J, Endre Z, Kellum JA, et al. Acute kidney
injury: an increasing global concern. Lancet 2013;382(9887):170–9.
[4] Palevsky PM, Molitoris BA, Okusa MD, Levin A, Waikar SS, Wald R, et al. Design of
clinical trials in acute kidney injury: report from an NIDDK workshop on trial meth-
odology. Clin J Am Soc Nephrol 2012;7(5):844–50.
[5] Mehta RL, Pascual MT, Soroko S, Chertow GM, Group PS. Diuretics, mortality, and
nonrecovery of renal function in acute renal failure. JAMA 2002;288(20):2547–53.
[6] Teixeira C, Garzotto F, Piccinni P, Brienza N, Iannuzzi M, Gramaticopolo S, et al. Fluid
balance and urine volume are independent predictors of mortality in acute kidney
injury. Crit Care 2013;17(1):R14.
[7] Uchino S, Doig GS, Bellomo R, Morimatsu H, Morgera S, Schetz M, et al. Diuretics and
mortality in acute renal failure. Crit Care Med 2004;32(8):1669–77.
[8] Bagshaw SM, Delaney A, Haase M, Ghali WA, Bellomo R. Loop diuretics in the man-
agement of acute renal failure: a systematic review and meta-analysis. Crit Care
Resusc 2007;9(1):60–8.
[9] Ho KM, Sheridan DJ. Meta-analysis of frusemide to prevent or treat acute renal fail-
ure. BMJ 2006;333(7565):420.
[10] Sampath S, Moran JL, Graham PL, Rockliff S, Bersten AD, Abrams KR. The efficacy of
loop diuretics in acute renal failure: assessment using Bayesian evidence synthesis
techniques. Crit Care Med 2007;35(11):2516–24.
[11] Redfors B, Bragadottir G, Sellgren J, Sward K, Ricksten SE. Acute renal failure is NOT
an “acute renal success”—a clinical study on the renal oxygen supply/demand rela-
tionship in acute kidney injury. Crit Care Med 2010;38(8):1695–701.
[12] De Torrente A, Miller PD, Cronin RE, Paulsin PE, Erickson AL, Schrier RW. Effects of
furosemide and acetylcholine in norepinephrine-induced acute renal failure. Am J
Physiol 1978;235(2):F131–136.
[13] Kramer HJ, Schuurmann J, Wassermann C, Dusing R. Prostaglandin-independent
protection by furosemide from oliguric ischemic renal failure in conscious rats. Kid-
ney Int 1980;17(4):455–64.
[14] Sward K, Valsson F, Sellgren J, Ricksten SE. Differential effects of human atrial natri-
uretic peptide and furosemide on glomerular filtration rate and renal oxygen con-
sumption in humans. Intensive Care Med 2005;31(1):79–85.
[15] Aravindan N, Aravindan S, Riedel BJ, Weng HR, Shaw AD. Furosemide prevents apo-
ptosis and associated gene expression in a rat model of surgical ischemic acute renal
failure. Ren Fail 2007;29(4):399–407.
[16] Aravindan N, Shaw A. Effect of furosemide infusion on renal hemodynamics and an-
giogenesis gene expression in acute renal ischemia/reperfusion. Ren Fail 2006;
28(1):25–35.
[17] Silbert BI, Ho KM, Lipman J, Roberts JA, Corcoran TB, Morgan DJ, et al. Does furose-
mide increase oxidative stress in acute kidney injury? Antioxid Redox Signal
2016;26(5):221–6.
[18] Cantarovich F, Rangoonwala B, Lorenz H, Verho M, Esnault VL. High-dose
flurosemide in acute renal failure study G: high-dose furosemide for established
ARF: a prospective, randomized, double-blind, placebo-controlled, multicenter
trial. Am J Kidney Dis 2004;44(3):402–9.
[19] Kleinknecht D, Ganeval D, Gonzalez-Duque LA, Fermanian J. Furosemide in acute
oliguric renal failure. A controlled trial. Nephron 1976;17(1):51–8.
[20] Shilliday IR, Quinn KJ, Allison ME. Loop diuretics in the management of acute renal
failure: a prospective, double-blind, placebo-controlled, randomized study. Nephrol
Dial Transplant 1997;12(12):2592–6.
[21] Vargas Hein O, Staegemann M, Wagner D, von Heymann C, Martin M, Morgera S,
et al. Torsemide versus furosemide after continuous renal replacement therapy
due to acute renal failure in cardiac surgery patients. Ren Fail 2005;27(4):385–92.
146 S.M. Bagshaw et al. / Journal of Critical Care 42 (2017) 138–146
[22] van der Voort PH, Boerma EC, Koopmans M, Zandberg M, de Ruiter J, Gerritsen RT,
et al. Furosemide does not improve renal recovery after hemofiltration for acute
renal failure in critically ill patients: a double blind randomized controlled trial.
Crit Care Med 2009;37(2):533–8.
[23] Faubel S, Chawla LS, Chertow GM, Goldstein SL, Jaber BL, Liu KD. Acute Kidney Injury
Advisory Group of the American Society of N: ongoing clinical trials in AKI. Clin J Am
Soc Nephrol 2012;7(5):861–73.
[24] Molitoris BA, Okusa MD, Palevsky PM, Chawla LS, Kaufman JS, Devarajan P, et al. Design
of clinical trials in AKI: a report from an NIDDK workshop. Trials of patients with sepsis
and in selected hospital settings. Clin J Am Soc Nephrol 2012;7(5):856–60.
[25] Bagshaw SM, Gibney RT, McAlister FA, Bellomo R. The SPARK Study: a phase II ran-
domized blinded controlled trial of the effect of furosemide in critically ill patients
with early acute kidney injury. Trials 2010;11:50.
[26] Ostermann M, Alvarez G, Sharpe MD, Martin CM. Frusemide administration in crit-
ically ill patients by continuous compared to bolus therapy. Nephron Clin Pract
2007;107(2):c70–76.
[27] Devine DJ. Gentimicin therapy. Drug Intell Clin Pharm 1974;8(11):650–5.
[28] Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P. Acute Dialysis Quality Initiative
w: Acute renal failure - definition, outcome measures, animal models, fluid therapy
and information technology needs: the Second International Consensus Conference
of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 2004;8(4):R204-12.
[29] Bone RC, Sprung CL, Sibbald WJ. Definitions for sepsis and organ failure. Crit Care
Med 1992;20(6):724–6.
[30] Hoste EA, Clermont G, Kersten A, Venkataraman R, Angus DC, De Bacquer D, et al.
RIFLE criteria for acute kidney injury are associated with hospital mortality in criti-
cally ill patients: a cohort analysis. Crit Care 2006;10(3):R73.
[31] Levi TM, Rocha MS, Almeida DN, Martins RT, Silva MG, Santana NC, et al. Furosemide
is associated with acute kidney injury in critically ill patients. Braz J Med Biol Res
2012;45(9):827–33.
[32] Nisula S, Kaukonen KM, Vaara ST, Korhonen AM, Poukkanen M, Karlsson S, et al. In-
cidence, risk factors and 90-day mortality of patients with acute kidney injury in
Finnish intensive care units: the FINNAKI study. Intensive Care Med 2013;39(3):
420–8.
[33] Wu VC, Lai CF, Shiao CC, Lin YF, Wu PC, Chao CT, et al. Effect of diuretic use on 30-day
postdialysis mortality in critically ill patients receiving acute dialysis. PLoS One
2012;7(3):e30836.
[34] Grams ME, Estrella MM, Coresh J, Brower RG, Liu KD, National Heart L. Blood Insti-
tute Acute Respiratory Distress Syndrome N: fluid balance, diuretic use, and mortal-
ity in acute kidney injury. Clin J Am Soc Nephrol 2011;6(5):966–73.
[35] Bouchard J, Soroko SB, Chertow GM, Himmelfarb J, Ikizler TA, Paganini EP, et al. Pro-
gram to Improve Care in Acute Renal Disease Study G: fluid accumulation, survival
and recovery of kidney function in critically ill patients with acute kidney injury.
Kidney Int 2009;76(4):422–7.
[36] Goldstein S, Bagshaw SM, Cecconi M, Okusa M, Wang H, Kellum J, et al. Pharmaco-
logical management of fluid overload. Br J Anaesth 2014;113(5):756–63.
[37] Gaudry S, Hajage D, Schortgen F, Martin-Lefevre L, Pons B, Boulet E, et al. Initiation
strategies for renal-replacement therapy in the intensive care unit. N Engl J Med
2016;375(2):122–33.
[38] Wald R, Adhikari NK, Smith OM, Weir MA, Pope K, Cohen A, et al. Comparison of
standard and accelerated initiation of renal replacement therapy in acute kidney in-
jury. Kidney Int 2015;88(4):897–904.
[39] Bagshaw SM, Delaney A, Jones D, Ronco C, Bellomo R. Diuretics in the management
of acute kidney injury: a multinational survey. Contrib Nephrol 2007;156:236–49.
[40] Jones SL, Martensson J, Glassford NJ, Eastwood GM, Bellomo R. Loop diuretic therapy
in the critically ill: a survey. Crit Care Resusc 2015;17(3):223–6.
[41] Chawla LS, Davison DL, Brasha-Mitchell E, Koyner JL, Arthur JM, Shaw AD, et al. De-
velopment and standardization of a furosemide stress test to predict the severity of
acute kidney injury. Crit Care 2013;17(5):R207.
[42] Silbert BI, Ho KM, Lipman J, Roberts JA, Corcoran TB, Morgan DJ, et al. Determinants
of urinary output response to IV furosemide in acute kidney injury: a pharmacoki-
netic/pharmacodynamic study. Crit Care Med 2016;44(10):e923-29.
[43] Wald R, Bagshaw SM. The timing of renal replacement therapy initiation in acute
kidney injury. Semin Nephrol 2016;36(1):78–84.