Feature Articles

Safety and Efficacy of Sodium Nitroprusside

During Prolonged Infusion in Pediatric Patients
Gregory B. Hammer, MD1,2; Andrew Lewandowski, PhD3; David R. Drover, MD1;
David A. Rosen, MD4; Carol Cohane, BSN, RN1; Ravinder Anand, PhD3; Jeff Mitchell, BS3;
Tammy Reece, MS, PMP, CCRA5; Scott R. Schulman, MD6

1
Department of Anesthesia, Stanford University School of Medicine,
­Stanford, CA.
2
Department of Pediatrics, Stanford University School of Medicine,
­Stanford, CA.
3
The EMMES Corporation, Rockville, MD.
4
Department of Anesthesiology, West Virginia University, Morgantown, WV.
5
Duke Clinical Research Institute, Durham, NC.
6
Department of Anesthesia, University of California-San Francisco School
of Medicine, San Francisco, CA.
The principal investigators of the coordinating and enrolling clinical cen-
ters were Gregory B. Hammer, MD (study coprincipal investigator); Scott
R. Schulman, MD (study coprincipal investigator); David A. Rosen, MD
(site principal investigator at West Virginia University Hospital, Morgan-
town, WV); Xiomara Garcia, MD (site principal investigator at Arkansas
Children’s Hospital, Little Rock, AR); Mary Hartman, MD (site principal
investigator at Duke University Medical Center, Durham, NC); Janice E.
Sullivan, MD (site principal investigator at the University of Louisville, Lou-
isville, KY); and Thomas P. Shanley, MD (site principal investigator at the
University of Michigan, Ann Arbor, MI).
Supplemental digital content is available for this article. Direct URL cita-
tions appear in the printed text and are provided in the HTML and PDF
versions of this article on the journal’s website (http://journals.lww.com/
pccmjournal).
Supported, in part, by the Eunice Kennedy Shriver National Institute of
Child Health and Human Development as part of the Best Pharmaceuti-
cals for Children Act.
Dr. Drover consulted for Johnson and Johnson, Inc. (legal); provided
expert testimony for Purdue (legal); and received support for article
research from the National Institutes of Health (NIH); and his institution
also received grant support from the NIH. Dr. Hammer received support
for article research from the NIH. His institution received grant support. Dr.
Lewandowski received support for article research from the NIH. His insti-
tution received support for travel and provision of materials and assistance
from the NIH. Dr. Rosen’s institution received grant support and support
for travel from the NICHD, served as a board member for Medicines Corp
(DSMB), consulted for REvo biologics, lectured for Maquet, and received
support for article research from the NIH. Dr. Cohane received support for
article research from the NIH; and her institution received grant support
from the NIH. Dr. Anand received support for article research from the
NIH. His institution received support for travel from the NICHD and the
NIH. Dr. Mitchell received support for article research from the NIH; and
his institution received support for travel and support for manuscript writ-
ing/review. Dr. Reece received support for article research from the NIH
and her institution received grant support from the NIH. Dr. Schulman has
disclosed that he does not have any potential conflicts of interest.
Objective: Sodium nitroprusside is a direct-acting vasodilator used
to lower blood pressure in the operating room and ICU. The efficacy
of sodium nitroprusside has been analyzed in few pediatric random-
ized trials. This study assesses the efficacy and safety of sodium
nitroprusside following at least 12 hours of IV infusion in children.
Design: Randomized, double-blind withdrawal to placebo study.
Setting: ICUs.
Patients: Pediatric patients younger than 17 years.
Interventions: Following 12–24 hours of open-label sodium nitro-
prusside titration, a blinded infusion of sodium nitroprusside or
placebo was administered (at the stable rate used at the end of
the open-label phase) for up to 30 minutes.
Measurements and Main Results: The primary efficacy measure
was whether control of mean arterial blood pressure was lost, that
is, increased above ambient baseline for two consecutive minutes
during the blinded phase. The proportion of patients who lost
mean arterial blood pressure control in the placebo group (15/19;
79%) was significantly different than those in the sodium nitroprus-
side group (9/20; 45%) (p = 0.048). Three patients experienced
rebound hypertension during the blinded phase, and all were in the
placebo group. Serious adverse event rates were low (7/52; 13%),
and in only one patient was the serious adverse event determined
to be related to sodium nitroprusside by the site investigator. Four-
teen patients (27%) had whole blood cyanide levels above 0.5 μg/
mL, with high correlation (0.7) between infusion rate and cyanide
levels, but there were few clinical signs of cyanide toxicity.
Conclusions: Sodium nitroprusside is efficacious in maintain-
ing mean arterial blood pressure control in children following a
12-hour infusion. Although a high proportion of patients were
found to have elevated cyanide levels, toxicity was not observed.
(Pediatr Crit Care Med 2015; 16:397–403)
Key Words: acidosis; antihypertensive agents; cyanide;
nitroprusside; pediatrics

For information regarding this article, E-mail: ddrover@stanford.edu
Copyright © 2015 by the Society of Critical Care Medicine and the World
Federation of Pediatric Intensive and Critical Care Societies
DOI: 10.1097/PCC.0000000000000383
B
lood pressure control in children is a significant con-
cern in the ICU, where management of hypertension
may be necessary during periods of acute physiologic
stress following surgical and medical procedures (1, 2). Sodium

Pediatric Critical Care Medicine www.pccmjournal.org 397

Copyright © 2015 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
 Hammer et al

nitroprusside (SNP) is a direct-acting vasodilator commonly prolonged infusion in pediatric patients. We analyzed the safety
used to control blood pressure in the ICU (3). Despite more and efficacy of SNP following 12 hours of infusion, with particular
than 50 years of widespread use in children for the treatment emphasis on the relationship between cyanide levels and toxicity.
of severe hypertension, there is limited information on the
safety and efficacy of SNP in children.
MATERIALS AND METHODS
SNP metabolism produces cyanide and thiocyanate. The
possibility of cyanide toxicity must be considered when Patients
administering SNP at high doses and/or for prolonged peri- Sixty-three patients treated at PICUs were enrolled in this
ods. In adults, the earliest, most sensitive signs of cyanide tox- study at five sites. Patients were younger than 17 years with
icity associated with SNP administration are acidosis, elevated weight at least 3 kg at baseline and were anticipated to need a
mixed venous oxygen tension, and rising blood lactate levels. reduction in mean arterial blood pressure (MAP) by at least
CNS manifestations of cyanide toxicity include headache, 20 mm Hg (≥ 15 mm Hg if < 2 yr old) from the baseline level
anxiety, agitation, confusion, lethargy, seizures, and coma. during open-label administration of at least 12 hours. Patients
Cardiovascular manifestations include progressive heart failure receiving stable doses of oral antihypertensive drugs prior
with both loss of cardiac muscle contractile force and slowing to the initiation of study SNP were eligible for enrollment.
of rate. However, prior studies have reported poor association Patients who received sodium thiosulfate within 6 hours of the
between elevated blood cyanide concentrations (≥ 0.5 μg/mL) start of open-label administration were excluded, although it
and signs of cyanide toxicity (4, 5). The authors concluded that was allowed after SNP initiation to treat suspected cyanide tox-
further pediatric studies were needed. icity. The supplemental data (Supplemental Digital Content 1,
The Best Pharmaceuticals for Children Act (BPCA) was signed http://links.lww.com/PCC/A149) contain complete inclusion/
into law to prioritize the study of off-patent drugs used in chil- exclusion criteria. Informed consent and, when applicable,
dren. This study (NCT00621816) and a pharmacokinetic/phar- assent were obtained prior to initiation of the study. Site insti-
macodynamic study of short-term SNP infusion (NCT00135668) tutional review boards reviewed the clinical protocol, protocol
were conducted under the BPCA to address the lack of adequate amendments, informed parental permission forms, and any
information about SNP dosing, pharmacokinetics, and safety in other form of patient information related to the study.
pediatric patients identified by the National Institutes of Health
in consultation with the Food and Drug Administration (FDA) Study Design
(6). We performed this multicenter, randomized, double-blind, As shown in the study design diagram in Figure 1, patients
placebo-controlled, parallel group study to determine whether were scheduled to receive open-label drug administration of
the hypotensive effect of SNP is lost and to assess the potential SNP for 12–24 hours. If the open-label phase began with study
for rebound hypertension following withdrawal to placebo after SNP, then SNP was initiated at up to 0.3 μg/kg/min and was
titrated to control MAP within
10% of a clinically determined
target level. Otherwise, admin-
istration of institutionally sup-
plied SNP for up to 8 hours was
allowed. The target MAP was
first determined at baseline and
was expected to start at least
15 or 20 mm Hg below MAPB1.
The target could be adjusted
during open-label phase SNP
administration. SNP adminis-
tration was halted if MAP fell
below 50 mm Hg (40 if < 1 mo
old). Infusion was restarted at a
lower rate once MAP increased.
Patients were randomized to
receive either SNP or placebo dur-
ing a subsequent double-blinded
study drug administration phase.
Randomization was stratified
by the 5-age groups shown in
Figure 2. Following at least 12
Figure 1. Scheduled study treatment phases with key mean arterial pressure (MAP) levels identified. SNP = hours of open-label SNP admin-
sodium nitroprusside. istration with SNP administered

398 www.pccmjournal.org June 2015 • Volume 16 • Number 5

Copyright © 2015 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
 Feature Articles

of treatment failure. Rebound

hypertension from pretreat-
ment levels was also compared
between the two blinded treat-
ment groups. Mild rebound
hypertension was defined by
MAP at least 110% of MAPB1.

Safety Assessments
Safety was assessed through
adverse events (AEs), seri-
ous adverse events (SAEs),
untoward changes in vital
signs (tachycardia and hypo-
tension), and laboratory
results (lactic acidosis and
elevated cyanide levels). AEs
were monitored and recorded
from SNP initiation up to 24
hours following study drug
discontinuation. SAEs were
recorded for 30 days follow-
Figure 2. Study flow diagram for patients who signed informed consent and enrolled into the study, received treat- ing discontinuation. Baseline
ment in either of the two study treatment phases, and were included in the primary efficacy analysis. SNP = sodium
nitroprusside, TS = tanner stage.
physical examinations were
recorded up to 7 days prior
to SNP initiation. Posttreatment physical examinations were
at a constant rate during the last 20 minutes, patients were eligible
to enter the blinded phase. Blinded infusion of SNP or placebo was performed within 2 hours and 18–30 hours after study drug
administered at the constant rate used at the end of the open-label discontinuation. The thresholds used to define these events
phase. New vasoactive drugs initiated during open-label infusion are presented in Supplemental Table 1 (Supplemental Digital
were recommended to be stopped at least three half-lives of the Content 1, http://links.lww.com/PCC/A149).
vasoactive drug prior to the start of the blinded phase. MAP was Open-label laboratory assessments for lactate, arterial blood
measured every minute during the 30-minute blinded phase. As gas (ABG), cyanide, and thiocyanate levels were collected every
shown in Figure 1, offset of MAP was defined by the blinded phase 8 hours (± 30 min) after SNP initiation. Additional collections
baseline MAP level (MAPB2) plus at least 50% of the difference were performed within 2 hours of study drug discontinuation
[Δ(MAP)] between MAPB2 and the open-label baseline MAP level and at follow-up (≤ 24 hr postdrug discontinuation). Samples
(MAPB1). The blinded phase was scheduled to end after either 30 for whole blood cyanide and serum thiocyanate were also
minutes of infusion or when blinded treatment was halted early collected 12 hours after SNP was discontinued if the arterial
due to a loss of MAP control (indicated by MAP ≥ the offset for catheter was still in place. Urine samples for thiocyanate were
two consecutive minutes), moderate rebound hypertension (≥ collected every 8 hours after discontinuation until the urinary
120% of MAPB1), or a safety concern. catheter was removed. Baseline laboratory measurements were
collected up to 7 days prior to SNP initiation, when possible.
Efficacy Assessments Patients were classified as having elevated cyanide levels if any
The persistence of the effect of SNP was primarily evaluated whole blood cyanide level after SNP initiation was at least 0.5
by determining whether the administration of SNP at a stable μg/mL. NMS Labs (Willow Grove, PA) analyzed cyanide sam-
dose during the blinded phase resulted in adequate MAP con- ples using spectrophotometry. The lower limit of quantification
trol. Patients were classified as blinded phase treatment failures (LLQ) was 0.05 μg/mL. Thiocyanate was analyzed at NMS Labs
if MAP control was lost, blinded drug infusion had to be ter- using an ion chromatography method with a LLQ of 1 μg/mL.
minated, or additional treatment was needed to maintain MAP Baseline laboratory values before SNP initiation were not
control. The proportion of treatment failures were compared available in all patients since some were formally enrolled in
between blinded treatment groups. the study after the start of SNP administration without predose
Patients were removed from the primary efficacy analysis study laboratory collections. Changes from baseline laboratory
if they had either Δ(MAP) = 0 or a stable blinded phase infu- values could not be calculated for these patients, although the
sion rate below the protocol limit of 0.5 μg/kg/min. Although change in condition was classified as not declining from base-
Δ(MAP) = 0 was not a protocol violation, MAP control could not line in the safety analysis if no abnormalities were found after
be adequately assessed in these participants using the definition SNP initiation.

Pediatric Critical Care Medicine www.pccmjournal.org 399

Copyright © 2015 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
 Hammer et al
Statistical Methods
One infant enrolled in the study twice. Data from the sec-
ond enrollment were excluded from all analyses. The safety
analysis population consisted of all other enrollments from
patients who received SNP. Open-label dosing and baseline
summaries used the safety analysis population. The blinded
phase failure rate was compared between the randomized
groups to determine the primary efficacy of SNP. Treat-
ment effect was adjusted for clinical site and age using logis-
tic regression. Sites with few patients (< 5) in the analysis
population were clustered together when assessing between
site variation. No other demographic or baseline variable
was determined to be sufficiently meaningful for inclusion
as a covariate in this analysis. p Values were not adjusted for
multiplicity as there was only one primary analysis. All other
analyses were exploratory.
Laboratory values were summarized using the worst value
recorded after SNP initiation except for cyanide analyses using
specific time points. The total SNP dose and average SNP infu-
sion rate, including blinded SNP administration, were corre-
lated with laboratory values for patients who completed the
open-label phase. Cyanide values below the LLQ (BLQ) were
imputed as 0.
Analyses used the following age groups unless noted oth-
erwise: younger than 6 years and 6 years old or older. Height
and weight z-scores were calculated using age and sex-specific
growth charts provided by the Centers for Disease Control
and Prevention. Quantitative summaries are presented as
the mean ± 1 sd or median (minimum, maximum) unless
otherwise specified. p Values for unadjusted group compari-
sons were evaluated using Fisher exact test for categorical
data and t tests with pooled variance for continuous data
unless noted otherwise. All CIs were calculated using 95%
confidence limits. Tests used a significance level of α = 0.05.
Analyses were conducted in SAS software version 9.3 (SAS
Institute, Cary, NC)
Dosing Summary (Mean ± sd) by Blinded Treatment Group in the Safety
Table 1.
Population
Blinded Treatment Group
Variable SNP (n = 24) Placebo (n = 21)
Open-label infusion duration (hr) 16.8 ± 4.1
Average open-label infusion rate 2.7 ± 1.7
(μg/kg/min)
Open-label dose (mg/kg) 2.7 ± 1.9
Study SNP infusion duration (hr) 17.2 ± 4.1
Average study SNP infusion rate 2.7 ± 1.7
(μg/kg/min)
Study SNP dose (mg/kg) 2.8 ± 1.9
Blinded infusion rate (μg/kg/min) 2.9 ± 2.3
SNP = sodium nitroprusside.
400 www.pccmjournal.org June 2015 • Volume 16 • Number 5
Copyright © 2015 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
RESULTS
Demographics, Dosing, and Baseline Characteristics
Fifty-one of the 63 enrolled patients (81%) received SNP and
were included in the safety analysis population. Forty-five of
the 51 safety analysis patients (88%) entered the blinded phase
after at least 12 hours of open-label dosing. Twenty-four of the
45 patients (53%) were randomized to receive blinded SNP.
Two additional randomized patients withdrew from the study
during the open-label phase following AEs of hypotension.
Three other early-terminated patients withdrew due to no lon-
ger needing treatment and one terminated due to an SAE that
began before SNP initiation.
The 30 days to less than 2-year group was the highest enroll-
ing age group in both blinded treatment groups and contained
39% of the 51 patients in the safety population (Fig. 2). In the
safety population, the median age was 3.8 years (0.03, 16.8).
Twenty-nine (57%) were male patients. Patients were enrolled
at five clinical sites. Thirty-five safety population patients (69%)
were enrolled at the University of West Virginia. Forty patients
(78%) started study SNP administration during or soon after a
surgical procedure. Most surgeries were cardiovascular, includ-
ing 12 for repair of coarctation of the aorta. Three patients had
reported noncardiovascular surgeries (G-tube placement, bron-
chus resection, and lumber fusion). Median height and weight
z-scores were slightly higher in the blinded SNP group (–0.6
and –0.4) than in the blinded placebo group (–1 and –0.8), and
both median z-scores were –0.7 in the overall safety population.
Median Δ(MAP) was similar in blinded SNP (28 mm Hg [0,
43 mm Hg]) and placebo (30 mm Hg [8, 63 mm Hg]) groups.
Open-label and overall study (open-label and blinded) SNP
administration is summarized in Table 1. Seventeen patients
(33%) received at least 3 mg/kg during the open-label phase. The
median total open-label dose was 2.3 mg/kg (0.6, 6) and 2.5 mg/
kg (0.1, 8.2) in the SNP and placebo groups, respectively. The
median blinded phase infusion rate was the same (2.5 μg/kg/
min) in both groups. For patients who completed the open-label
Early Terminated (n = 6) Total (n = 51)
15.9 ± 4.1 2.7 ± 2.1 14.7 ± 5.9
3.1 ± 1.6 0.7 ± 0.3 2.6 ± 1.7
3.1 ± 2.1 0.1 ± 0.1 2.6 ± 2.1
15.9 ± 4.1 2.7 ± 2.1 15 ± 6
3.1 ± 1.6 0.7 ± 0.3 2.6 ± 1.7
3.1 ± 2.1 0.1 ± 0.1 2.6 ± 2.1
3.5 ± 2.6 Not applicable 3.2 ± 2.4
 Feature Articles

phase, the mean average open-label infusion rate was 3.27 ± 1.7
μg/kg/min (median, 3.38) in patients younger than 6 (n = 28)
and 2.28 ± 1.42 μg/kg/min (median, 1.91) in older patients (n
= 17). The difference in mean average open-label infusion rate
between age groups was 0.99 μg/kg/min with 95% CI (0.04, 2) (p
= 0.043). The Pearson correlation between age and average infu-
sion rate was –0.33 (p = 0.026). Median infusion rate was lower
in each successive age group (Supplemental Fig. 1, Supplemental
Digital Content 1, http://links.lww.com/PCC/A149).
Efficacy Analysis
Twenty of the 24 patients (83%) in the blinded SNP group and
19 of the 21 patients (90%) in the blinded placebo group were
included in the primary efficacy analysis. In the SNP group, two
patients were removed since Δ(MAP) = 0. Also, one patient in the
placebo group and two patients in the SNP group were removed
due to a dose less than 0.5 μg/kg/min, and another patient in the
placebo group had no MAP measurements in the first 28 min-
utes of the blinded phase. Fifteen of the 19 patients in the pla-
cebo (79%) and only nine of the 20 patients in the SNP (45%)
groups experienced treatment failure (p = 0.048). The odds of
treatment failure were 4.6 times higher in the placebo group
with the 95% CI (1.1, 18.8). When adjusted for site and age, the
estimated odds ratio was 5 (1.3, 24.6). The adjusted model had a
significant treatment effect (p = 0.03), a concordance coefficient
of 0.75, and nonsignificant effects for age (p = 0.44) and site (p =
0.43). Failure times were also significantly different between the
two groups (Supplemental Fig. 2, Supplemental Digital Content
1, http://links.lww.com/PCC/A149). MAP values in the two SNP
group patients with Δ(MAP) = 0 were stable during the blinded
phase with all values less than 10% from MAPB2.
Three of the 21 patients (14%) in the placebo group and no
patients in the SNP group experienced mild rebound hyper-
tension during the blinded phase. No patient experienced
moderate rebound hypertension.
Safety Analysis
Eight SAEs were experienced by seven of the 51 patients (14%)
in the safety population. No SAEs were experienced by patients
in the oldest age group. Two SAEs, sepsis and respiratory fail-
ure, led to death. The deaths occurred 2 and 30 days after SNP
administration had stopped, and neither death was reported as
being related to study treatment. The two related SAEs (blood
cyanide increase and respiratory depression) were reported in
the same patient. This patient experienced no SAEs and one
unrelated AE (pyrexia) during the second enrollment. One
unrelated SAE, reported as hemorrhage, led to SNP discon-
tinuation but was first discovered before SNP initiation. The
other unrelated SAEs were apnea, cerebral spinal fluid (CSF)
leak, and ventricular fibrillation.
Thirty-one patients (61%) experienced 93 AEs, including
the eight SAEs. Thirty of these AEs, experienced by 16 patients
(31%), were classified as related to SNP. In addition to the two
related SAEs, five AEs were classified as being both of moderate
severity and at least possibly related to treatment. Six patients
(12%) were classified as having tachycardia. No relationship
between age and AE occurrence was found.
Thirty-one patients (60%) experienced hypotension as
defined by a drop in MAP more than 20% below the target
at some point during the open-label or blinded phase. This
includes both patients who terminated during the open-label
phase following hypotension AEs. Only one patient (2%) met
all criteria for severe hypotension requiring pharmacological
therapy (Supplemental Table 1, Supplemental Digital Content 1,
http://links.lww.com/PCC/A149), but six other patients (12%)
had a recorded MAP value more than 30% below the target.
Fourteen patients (27%) had elevated whole blood cyanide
levels (≥ 0.5 μg/mL). Elevated levels were most common in
the less than 30-day (2/3; 67%) and 30-day to less than 2-year
(8/20; 38%) age groups. Twelve of these patients had levels more
than 1.5 μg/mL. Baseline blood cyanide values were BLQ in all
patients with baseline measurements. Two patients received

Table 2. Summaries (Mean or Median) and Correlations With Study Dose (mg/kg) and
Average Study Infusion Rate (μg/kg/min) for Clinical Laboratory Assessment Changes
From Baseline
Change From Baseline Correlations

Variable n Summary n Dose Rate

Blood cyanide (μg/mL) 50 1 ± 1.9 44 0.66 0.71

Blood thiocyanate (μg/mL) 45 6.5 ± 4.9 39 0.58 0.56
Urine thiocyanate (μg/mL) 40 4 ± 3.5 34 0.4 0.33
Lactate (mmol/L) 45 0.2 ± 1.2 41 0.57 0.52
HCO3 (mEq/L) 48 –0.2 ± 2.4 43 –0.18 –0.19
pH 48 0.01 ± 0.08 43 0.09 0.02
Creatinine (mg/dL) 48 0 (–0.7, 0.3) 43 0.23 0.24
Alanine transaminase (U/L) 49 –3 (–333, 63) 45 –0.17 –0.17
Aspartate transaminase (U/L) 48 –3.5 (–1,016, 79) 44 –0.32 –0.29

Pediatric Critical Care Medicine www.pccmjournal.org 401

Copyright © 2015 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
 Hammer et al
sodium thiosulfate after the end of study SNP administra-
tion. Only one of these administrations was indicated as being
administered as a cyanide antidote, but this was the only patient
without an analyzable postbaseline whole blood cyanide sam-
ple. High correlations were found between the greatest postdose
cyanide level and both average study SNP infusion rate and total
dose (Table 2). Also, no cyanide levels at least 0.5 μg/mL were
found in any patient with an average study infusion rate less
than 3 μg/kg/min (Supplemental Fig. 3, Supplemental Digital
Content 1, http://links.lww.com/PCC/A149).
Supplement Table 2 (Supplemental Digital Content 1,
http://links.lww.com/PCC/A149) summarizes the 8 hours
postdose, study drug discontinuation, and greatest study
cyanide levels by age group (< 6 and ≥ 6 yr). At the 8-hour
time point, all patients with average infusion rates less than
3 μg/kg/min, which is equivalent to less than 1.44 mg/kg over
8 hours, also had BLQ cyanide levels (Supplemental Fig. 4,
Supplemental Digital Content 1, http://links.lww.com/PCC/
A149). The Spearman correlations were high between infu-
sion rates and cyanide levels through both 8 hours (0.69; n =
44) and study discontinuation (0.7; n = 39). Patients younger
than 6 years had higher rates of elevated cyanide levels, but
these rates were not significantly different between the two age
groups (p = 0.11 for the worst study cyanide level). Four of the
14 patients (29%) with elevated cyanide levels still had elevated
levels in follow-up measurements taken at least 11 hours after
study SNP discontinuation, although two were still on stan-
dard of care SNP at the time of the last cyanide measurement.
Of the 14 patients with elevated cyanide levels, 11 (79%)
had recorded AEs. In the 36 patients with evaluable blood cya-
nide levels less than 0.5 μg/mL, the proportion (53%) with AEs
was lower but not significantly different (p = 0.12). SAE rate
was also higher in the group with elevated cyanide (21% vs
11%) but not significantly different (p = 0.38). SAEs recorded
in patients with high cyanide levels were CSF leak, respiratory
depression, respiratory failure, and an increase in blood cya-
nide levels. The patient with both respiratory depression and
increase in blood cyanide SAEs also had the highest blood cya-
nide level recorded in the study (6.5 μg/mL). One death was
reported in each group. The death due to respiratory failure
was in a patient whose greatest cyanide level was 3.8 μg/mL.
Patients with (3; 21%) and without (6; 17 %) elevated cyanide
levels had similar rates of gastrointestinal AEs. No patient with
elevated cyanide levels had a cardiac disorder commonly asso-
ciated with cyanide toxicity.
Additional laboratory values of interest are summarized in
Table 2, which includes correlations with study (open-label and
blinded) SNP dose and infusion rate. Blood thiocyanate val-
ues were not severely high, although the correlations between
dosing summaries and change scores were more than 0.5. The
two highest urine thiocyanate levels of 44 and 22 μg/mL were
found in patients without valid baseline measurements. Only
one patient met the 5 mmol/L threshold for lactic acidosis (5.5
mmol/L) after SNP initiation, although this patient’s lactate
level was higher (5.8 mmol/L) at baseline. Thirteen of the 49
patients (27%) with postbaseline data had recorded lactate
402 www.pccmjournal.org June 2015 • Volume 16 • Number 5
Copyright © 2015 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
levels between 2.5 and 5 mmol/L. Seven of these 13 patients
also had baseline values above 2.5 mmol/L, and two patients
had no recorded baseline value. Median postdose pH (7.39)
and HCO3 (25 mEq/L) are in the normal range (Supplemental
Table 3, Supplemental Digital Content 1, http://links.lww.
com/PCC/A149). Mean change scores for both ABG labora-
tory values are near 0 with low variability and low correlation
with dosing measures. Two patients received sodium bicarbon-
ate after SNP initiation. Five patients experienced creatinine
increases or decreases at least 0.3 mg/dL with no relationship
between creatinine change and cyanide levels (Supplemental
Fig. 5, Supplemental Digital Content 1, http://links.lww.com/
PCC/A149).
DISCUSSION
We performed a randomized, double-blind withdrawal to
placebo study of SNP in children following 12–24 hours of
open-label SNP infusion. Tolerance was defined as no change
in MAP during administration of placebo. The median SNP
infusion rate across both study phases was 2.7 μg/kg/min over
a median duration of 14.5 hours. As previously reported (7,
8), we found that younger patients (< 6 yr old) needed higher
infusion rates than older patients. We also found high correla-
tions between cyanide levels and both infusion rate and total
dose. This finding is supported by results of a prior study by
Moffett and Price (5) that suggested an average SNP infusion
rate of 1.8 μg/kg/min as a threshold for predicting cyanide lev-
els at least 0.5 μg/mL in children following cardiac surgery. The
authors reported an 11% incidence rate of elevated cyanide
levels compared with 27% in our study.
Przybylo et al (9) described cyanide and thiocyanate blood
levels in 10 children who received SNP at doses up to 10 μg/kg/
min (with a mean infusion rate of 6 μg/kg/min) while undergo-
ing cardiopulmonary bypass for repair of complex congenital
cardiac defects. Although some children had whole blood cya-
nide levels at least 0.5 μg/mL, no patient developed clinically
apparent toxicity. As was also reported in a literature review
of past studies using SNP (10), we did not find a relationship
between cyanide elevation and cyanide-related toxicity.
The SAEs reported in our study were unrelated to cyanide
toxicity. Rates of AEs (including gastrointestinal problems and
cardiac events associated with cyanide) were comparable in
patients with and without elevated cyanide levels. Twelve of
the 14 patients (86%) with detected cyanide levels had severely
elevated levels more than 1.5 μg/mL. The AE rate was higher
in patients with elevated cyanide levels, but AEs commonly
related to cyanide toxicity were either not reported in the
study population or occurred at similar rates in patients with
or without elevated cyanide levels. AEs were recorded for 24
hours and SAEs were recorded for 30 days after study SNP dis-
continuation. Thus, some AEs associated with cyanide toxicity
that had a late onset may not have been captured.

CONCLUSIONS
In this withdrawal to placebo study, we evaluated offset of
blood pressure control (tolerance) during SNP infusion in
children following 12–24 hours of open-label infusion. Our
analysis showed a significant difference in the rates of MAP
increase during the placebo-controlled blinded infusion
period in patients receiving SNP versus placebo. The odds of
maintaining blood pressure control were 4.6 times higher in
patients receiving SNP. In addition, all three cases of rebound
hypertension during the blinded phase were found in the pla-
cebo group. SNP is effective in maintaining blood pressure
control in the pediatric study population without the develop-
ment of tolerance during 12 hours of infusion.
Aside from elevated cyanide levels, we found few safety
concerns related to SNP. Patients in this study showed little
change in lactate levels or kidney function, few indications
of metabolic acidosis, and no seriously elevated blood thio-
cyanate levels. Two patients withdrew during the open-label
phase following AEs of hypotension. Although high cya-
nide levels were detected in patients on higher doses and
may need to be monitored in practice, SNP was found to
be safe in children at the doses used in this study. Efficacy
and safety findings from this study and study NCT00135668
were reviewed by the FDA and used to revise FDA labeling of
SNP for pediatric use (6).
ACKNOWLEDGMENTS
We thank Michelle Dulik (site coordinator at West Vir-
ginia University) for her contributions to data collection.
Pediatric Critical Care Medicine www.pccmjournal.org 403
Copyright © 2015 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
Feature Articles
We thank Anne Zajicek (PharmD, MD; Eunice Kennedy
Shriver National Institute of Child Health and Human
Development) for her contributions to study design and
implementation.
REFERENCES
1. Groshong T: Hypertensive crisis in children. Pediatr Ann 1996;
25:368–371, 375–376.
2. Sinaiko AR: Hypertension in children. N Engl J Med 1996;
335:1968–1973
3. Tinker JH, Michenfelder JD: Sodium nitroprusside: Pharmacology,
toxicology and therapeutics. Anesthesiology 1976; 45:340–354
4. Linakis JG, Lacouture PG, Woolf A: Monitoring cyanide and thiocya-
nate concentrations during infusion of sodium nitroprusside in chil-
dren. Pediatr Cardiol 1991; 12:214–218
5. Moffett BS, Price JF: Evaluation of sodium nitroprusside toxic-
ity in pediatric cardiac surgical patients. Ann Pharmacother 2008;
42:1600–1604
6. Pediatric Studies of Sodium Nitroprusside Conducted in Accordance
With the Public Health Service Act; Availability of Summary Report
and Requested Labeling Changes. 79 Fed Reg 4167, Office of the
Federal Register, National Archives and Records Administration,
2014. Available at: http://www.gpo.gov/fdsys/granule/FR-2014-01-
24/2014-01390. Accessed February 9, 2014
7. Davies DW, Kadar D, Steward DJ, et al: A sudden death associated
with the use of sodium nitroprusside for induction of hypotension dur-
ing anaesthesia. Can Anaesth Soc J 1975; 22:547–552
8. Bennett NR, Abbott TR: The use of sodium nitroprusside in children.
Anaesthesia 1977; 32:456–463
9. Przybylo HJ, Stevenson GW, Schanbacher P, et al: Sodium nitroprus-
side metabolism in children during hypothermic cardiopulmonary
bypass. Anesth Analg 1995; 81:952–956
10. Thomas C, Svehla L, Moffett BS: Sodium-nitroprusside-induced
cyanide toxicity in pediatric patients. Expert Opin Drug Saf 2009;
8:599–602