Anda di halaman 1dari 7

Pediatric Anesthesia ISSN 1155-5645


Development of an optimal sampling schedule for children

receiving ketamine for short-term procedural sedation and
Catherine M.T. Sherwin1, Chris Stockmann1, Kristin Grimsrud2, David W. Herd3, Brian J. Anderson4
& Michael G. Spigarelli1
1 Division of Clinical Pharmacology, Department of Paediatrics, University of Utah School of Medicine, Salt Lake City, UT, USA
2 Campus Veterinary Services and School of Veterinary Medicine, University of California, Davis, CA, USA
3 Department of Anaesthesiology, University of Auckland, Auckland, New Zealand
4 Paediatric Emergency Department, Mater Children’s Hospital, Brisbane, Australia

Keywords Summary
optimal sampling strategy;
pharmacokinetics; pharmacodynamics; Background: Intravenous racemic ketamine is commonly administered for
children procedural sedation, although few pharmacokinetic studies have been con-
ducted among children. Moreover, an optimal sampling schedule has not
Correspondence been derived to enable the conduct of pharmacokinetic studies that minimally
Catherine M.T. Sherwin, PhD, FCP, Division
inconvenience study participants.
of Clinical Pharmacology, Department of
Methods: Concentration-time data were obtained from 57 children who
Paediatrics, University of Utah School of
Medicine, Salt Lake City, UT 84108, USA received 1–1.5 mgkg1 of racemic ketamine as an intravenous bolus. A pop-
Email: ulation pharmacokinetic analysis was conducted using nonlinear mixed
effects models, and the results were used as inputs to develop a D-optimal
Section Editor: Jerrold Lerman sampling schedule.
Results: The pharmacokinetics of ketamine were described using a two-com-
Accepted 5 July 2014
partment model. The volume of distribution in the central and peripheral
compartments were 20.5 l∙70 kg1 and 220 l∙70 kg1, respectively. The inter-
compartmental clearance and total body clearance were 87.3 and
87.9 lh1∙70 kg1, respectively. Population parameter variability ranged
from 34% to 98%. Initially, blood samples were drawn on 3–6 occasions
spanning a range of 14–152 min after dosing. Using these data, we deter-
mined that four optimal sampling windows occur at 1–5, 5.5–7.5, 10–20, and
90–180 min after dosing. Monte Carlo simulations indicated that these sam-
pling windows produced precise and unbiased ketamine pharmacokinetic
parameter estimates.
Conclusion: An optimal sampling schedule was developed that allowed
assessment of the pharmacokinetic parameters of ketamine among children
requiring short-term procedural sedation.

a neuro-protective effect for patients with traumatic

brain injuries (3); and in children, it has been used for
Ketamine is an N-methyl-D-aspartate (NMDA) antago- conscious sedation in the intensive care unit and
nist and is widely used for sedation, induction, and emergency department (4).
maintenance of general anesthesia and analgesia (1). There is a scarcity of pediatric ketamine
The benefits of ketamine have been well documented in pharmacokinetic data available (5–7). Nevertheless,
special populations. In burn patients, ketamine has been the use of ketamine in pediatric emergency depart-
reported to improve analgesia and reduce the need for ments has increased over the last three decades and
opioids (2); in combat trauma, it has been shown to have it is now commonly administered intra-operatively

© 2014 John Wiley & Sons Ltd 211

Pediatric Anesthesia 25 (2015) 211–216
Optimal ketamine sampling schedule C.M.T. Sherwin et al.

and postoperatively for pain control (8–11). The Covariate selection was performed using a stepwise
primary indications for the use of ketamine in the forward addition and backward elimination procedure
pediatric emergency department setting include that has been described previously (19). Age and
fracture reduction, joint relocation, facial laceration height were examined in the covariate analysis to
repair, wound care and cleaning, incision and drain- assess their influence upon ketamine clearance, the
age of abscesses, foreign body removal, and burn volume of distribution in the central compartment,
debridement and dressing (12,13). Despite its frequent and the volume of distribution in the peripheral com-
use, ketamine dosing protocols vary widely across partment. The median age of the 57 children included
institutions (e.g., single bolus vs multiple dosing, in the study was 8 (interquartile range [IQR] 5.6–
co-administration with benzodiazepines or an antisi- 9.4) years. The median weight of the patients included
alogogue, and the rate of drug administration) (14). in this study was 25.4 (IQR 21.6–40.3) kg. The med-
We previously reported significant variability in ian height was 129 (IQR 115.0–142.0) cm. There were
measured ketamine concentrations, despite administer- no neonates and an insufficient number of infants
ing dosages that were normalized per kilogram of body enrolled to evaluate the influence of maturation on
weight (15). Moreover, the relationship between keta- ketamine pharmacokinetics.
mine dose, plasma concentrations, and analgesic effects Ketamine pharmacokinetic parameter estimates were
is poorly defined in children (7,16). Consequently, we standardized for a mean body weight of 70 kg using an
sought to develop an optimal sampling schedule to allometric model of the form:
accurately characterize the pharmacokinetics of keta-  
mine among a cohort of children requiring short-term Wi PWR
Pi ¼ Pstd  ;
sedation. The primary objective was to provide blood Wstd
sampling guidance for future studies. where Pi is the pharmacokinetic parameter for the i th
participant, Wi is the weight of the i th participant, and
Methods Pstd is the parameter in a theoretical participant with a
Wstd of 70 kg. This standardization facilitates compari-
Clinical data sons of pediatric parameter estimates with those
reported for adult subjects. The PWR exponent was
Clinical data were taken from a published study (15). fixed at 0.75 for clearance and 1.0 for the central and
All children who received ketamine sedation at the peripheral volumes of distribution (20,21).
Auckland Children’s Hospital Emergency Department The Bayesian information criterion (BIC) and the
were approached for participation in this study during a objective function value were used to perform model
6-month period in 2006. The Regional Ethics Commit- selection among non-nested and nested models, respec-
tee reviewed and approved this study. Parental permis- tively (18,22,23). Additionally, visual examination of the
sion and assent (when applicable) were obtained before diagnostic plots and prediction-corrected visual predic-
conducting any study-related procedures. Children tive checks was performed to evaluate the fit of the phar-
undergoing painful procedures received a bolus dose of macokinetic model to the data.
racemic ketamine at 1–1.5 mgkg1 intravenously. Three
to six venous blood samples were obtained from each Identification of optimal sampling times
participant over a period of 14–152 min postketamine
administration. Serum concentrations were established An optimal sampling schedule was developed using pop-
using a validated method that has been previously ulation pharmacokinetic parameter estimates as a priori
described (17). values. The selection of optimal sampling points was
established according to the D-optimality algorithm
implemented in the software program ADAPT 5 (24). D-
Development of a population pharmacokinetic model optimal designs minimize the determinant of the inverse
A population pharmacokinetic model was developed Fisher information matrix and minimize the overall vari-
for ketamine using the nonlinear mixed effects ance across all pharmacokinetic parameter estimates
modelling software NONMEM 7.2 (ICON Development (25). D-optimal design theory explicitly assumes that
Solutions, Ellicott City, MD, USA) (18). Parameters there is one true vector for each pharmacokinetic param-
were estimated using the first-order conditional eter, which results in the number of optimal samples
estimation with interaction (FOCEI) algorithm. The being exactly equal to the number of pharmacokinetic
population parameter variability was ascribed to an parameters to be estimated (26). In this study, a two-
exponential model. compartment model was used, which resulted in a total

212 © 2014 John Wiley & Sons Ltd

Pediatric Anesthesia 25 (2015) 211–216
C.M.T. Sherwin et al. Optimal ketamine sampling schedule

of four pharmacokinetic parameters. Consequently, four basis for this approach has been well described by Foo
optimal sampling times were identified; however, alterna- and Duffull (28).
tive designs involving the collection of five and six blood
samples were also explored and their results are pre-
sented below.
Ketamine is commonly used in the emergency depart-
Population pharmacokinetic model
ment; consequently, a sampling window of 3 h was used
so as not to result in excessively protracted emergency Serum samples were available from 57 children for a
department visits. The validity of the sampling schedule total of 210 unique specimens. The pharmacokinetic
was evaluated using Monte Carlo techniques in which a parameters of ketamine were described with a two-com-
trial of 57 children with a median weight of 25.4 (IQR partment model (Table 1). A prediction-corrected visual
21.6–40.3) kg was simulated. Individual pharmacoki- predictive check was also performed to ensure that the
netic parameters were developed for this simulated
patient population, which were used to simulate keta-
mine concentrations at each of the optimal sampling
times. A new population model was then developed
based upon the concentrations for these simulated
patients. Individual Bayesian estimates were obtained
for each patient and compared with the original phar-
macokinetic parameters, which facilitated calculation of
the bias and precision of the estimated pharmacokinetic
parameters using the optimal sampling schedule (27).
This procedure was performed 2000 times to obtain
mean (standard deviation [SD]) bias and precision

Sampling windows
As clinical and logistical circumstances often make it
challenging to obtain a blood sample at a precisely
defined time, we developed sampling windows around Figure 1 Prediction-corrected visual predictive check of the final
each of the optimal sampling points to allow for this population ketamine model. The increase in ketamine concentrations
design to be applied in future prospective clinical observed at approximately 60 min was due to the administration of
studies. The sampling windows were derived using an additional dose. The solid gray line indicates the median, and
WinPOPT, which allowed us to identify a window of dashed gray lines indicate the 10th and 90th percentiles of the
observed data, respectively. The solid black line indicates the med-
time in which a nearly optimal blood sample could be
ian, and dashed black lines indicate the 10th and 90th percentiles of
obtained. We specified that each sampling window was the simulated data, respectively. Open circles depict the prediction-
to be constrained in time to achieve 90% efficiency in corrected observed concentrations. Predictions are aggregated
pharmacokinetic parameter estimation. The Bayesian across all 100 simulated trials.

Table 1 Ketamine population pharmacokinetic parameter and residual error estimates

Bootstrapped 95% % Population parameter %

Parameter Units Estimate confidence interval variability Shrinkage

Pharmacokinetic parameters
CL l h1∙70 kg1 87.3 64.8–111 52.9 9.7
Vcentral l∙70 kg1 20.5 13.4–33.1 97.8 7.4
Q l h1∙70 kg1 220 166–217 60.7 9.0
Vperipheral l∙70 kg1 87.7 57.9–123 34.1 30.4
Residual error
Additive – 0 FIXED – – –
Proportional % 21.2 16.2–25.8 – 21.2

© 2014 John Wiley & Sons Ltd 213

Pediatric Anesthesia 25 (2015) 211–216
Optimal ketamine sampling schedule C.M.T. Sherwin et al.

model predictions matched the measured ketamine con- Sampling times in children are not always possible at
centration-time course (Figure 1). After accounting for precisely defined optimal time points because of the
the allometric scaling of weight, neither of the covariates child’s clinical instability, noncompliance, or coincident
tested exerted a significant influence upon ketamine blood sampling for other medical reasons.
pharmacokinetic parameter estimates. Consequently, sampling windows were developed using
Bayesian methods to identify windows of opportunity
around each sampling point to achieve 90% precision in
Optimal sampling times
estimating ketamine pharmacokinetic parameters. The
Use of the D-optimality algorithm yielded an optimal four samples may be taken at 1–5, 5.5–7.5, 10–20, and
sampling schedule with blood draws at 1, 5.5, 16, and 90–180 min postdose. It must be noted that the exact
180 min postdose (Figure 2). By collecting a single times must be recorded when each of the samples are
sample at each of these time points, the coefficient of obtained to facilitate accurate population pharmacoki-
variation (%CV) for each of the pharmacokinetic netic parameter estimation.
parameters is expected to range from 13% to 39%.
Alternative designs involving the collection of five or six
samples and their respective %CV and optimal sampling
times are presented in Table 2. This study establishes an optimal sampling schedule for
The predictive performance of the optimal sampling the assessment of ketamine pharmacokinetic parameters
schedule involving the collection of four blood samples in children. Several techniques and software platforms
was evaluated using Monte Carlo simulation techniques have been developed to identify the most informative
(n = 2000). The bias and precision of the four ketamine times to obtain serial pharmacokinetic samples accord-
pharmacokinetic parameters are featured in Table 3. ing to optimal sampling theories (29–31). In this study,
The estimates obtained for all four pharmacokinetic we used a Bayesian approach that estimates the full
parameters had bias <11% and precision within 41%. pharmacokinetic profile for each individual using a lim-
ited number of ketamine plasma concentration measure-
ments over a period of 3 h (32). Four samples collected
at 1–5, 5.5–7.5, 10–20, and 90–180 min were found to
adequately characterize ketamine pharmacokinetic
parameters with no significant bias and precision rang-
ing from 8% to 41%.
Controlling pain and minimizing stressful proce-
dures is vitally important in critically ill pediatric
patients. Ketamine is frequently used in the emer-
gency department to provide short-term sedation and
analgesia for children undergoing minor surgical
procedures (14). Successful procedural sedation is a
direct function of administering the optimal dose at
the ideal dosing interval (33). However, rational
dosing of medications in pediatric patients remains
an unmet clinical need. Our limited understanding of
ketamine pharmacokinetics in children has historically
resulted in initial dosing regimens that vary greatly
(8,10,34). The availability of several large case series
has helped improve clinical dosing decisions with
regard to patient safety, although these were not
designed to enhance therapeutic effectiveness at the
level of the individual patient (35,36). In this study, we
adopted a Bayesian approach to allow for the identifi-
cation of the most ‘information-rich’ areas of the drug
concentration-time curve using the least number of
Figure 2 Optimal sampling strategy to accurately estimate the pop- blood samples (27,37). Using Monte Carlo simulation,
ulation pharmacokinetics of ketamine among children requiring short- these sampling times were found to be unbiased and
term sedation in the emergency department. reasonably precise, which suggests that they may be

214 © 2014 John Wiley & Sons Ltd

Pediatric Anesthesia 25 (2015) 211–216
C.M.T. Sherwin et al. Optimal ketamine sampling schedule

Table 2 Ketamine optimal sampling designs involving the collection of 4, 5, and 6 blood samples

Expected % coefficient
Pharmacokinetic parameter Parameter estimate of variation

4 sample design
Total body clearance (CL) 87.9 lh1∙70 kg1 13.2
Volume of distribution in the central compartment (Vcentral) 20.5 l∙70 kg1 20.0
Intercompartmental clearance (Q) 87.3 lh1∙70 kg1 39.3
Volume of distribution in the peripheral compartment (Vperipheral) 220 l∙70 kg1 18.4
Optimal sampling times 1, 5.5, 16, and 180 min
5 sample design
Total body clearance (CL) 87.9 lh1∙70 kg1 10.1
Volume of distribution in the central compartment (Vcentral) 20.5 l∙70 kg1 20.0
Intercompartmental clearance (Q) 87.3 lh1∙70 kg1 37.7
Volume of distribution in the peripheral compartment (Vperipheral) 220 l∙70 kg1 15.2
Optimal sampling times 1, 5.5, 16 (n = 2), and 180 min
6 sample design
Total body clearance (CL) 87.9 lh1∙70 kg1 8.7
Volume of distribution in the central compartment (Vcentral) 20.5 l∙70 kg1 17.9
Intercompartmental clearance (Q) 87.3 lh1∙70 kg1 37.1
Volume of distribution in the peripheral compartment (Vperipheral) 220 l∙70 kg1 14.9
Optimal sampling times 1, 5.5, 16 (n = 3), and 180 min

Table 3 Bias and precision of ketamine pharmacokinetics deter- do not completely describe the entire pharmacokinetic
mined through Monte Carlo simulation profile of the drug (39). However, incomplete data,
Mean % Bias Mean % Precision model misspecification, and high population parameter
Parameter (SD) (SD) variability may decrease the sensitivity of our optimal
sampling strategy.
CL 0.8 (10.2) 8.2 (6.1)
Vcentral 10.5 (61.3) 39.1 (48.2)
Q 1.1 (16.5) 13.0 (10.2) Conclusion
Vperipheral 10.6 (55.3) 40.9 (38.5)
The data generated from this study will provide a
framework for improving study designs of forthcom-
used prospectively to clearly define the population ing pediatric ketamine studies and minimize sample
pharmacokinetics of ketamine among children undergo- collection in this vulnerable patient population.
ing short-term sedation in the emergency department Future studies to characterize the pharmacokinetics
using a limited number of samples. and pharmacodynamics of ketamine in special popu-
Pharmacokinetic and pharmacodynamic studies are lations are warranted to optimize dosing regimens
difficult to perform in children due to ethical and hereby improving patient care and minimizing the
logistical constraints, including the limited number and incidence of adverse events.
quantity of specimens, as well as the extent and inva-
siveness of study-related interventions that are not part
of routine clinical care (38). To overcome some of
these challenges, we have developed a D-optimal phar- Financial support was provided by research grants from
macokinetic study design. Optimal sampling theory the Society for Paediatric Anaesthesia of New Zealand
requires a priori knowledge of the pharmacokinetic and Australia (SPANZA). Dr. Herd was supported by
parameters to be estimated. In this study, we devel- the Royal Australasian College of Physicians, the Mau-
oped a population pharmacokinetic model using data rice and Phyllis Paykel Trust, and the Joan Mary Rey-
from 57 children who each contributed 3–6 ketamine nolds Fellowship.
concentrations. These population pharmacokinetic
parameter estimates were then incorporated within a
Ethical approval
Bayesian model to estimate individual pharmacokinetic
parameters. This approach enhances the accuracy and This study was approved by the Regional Ethics Com-
applicability of estimates obtained from datasets that mittee and written informed consent was obtained from

© 2014 John Wiley & Sons Ltd 215

Pediatric Anesthesia 25 (2015) 211–216
Optimal ketamine sampling schedule C.M.T. Sherwin et al.

parents/guardians. Additionally, assent was obtained Conflict of interest

from children 6 years of age and older.
No conflicts of interest declared.

1 Persson J. Wherefore ketamine? Curr Opin emergency department paediatric procedural pharmacokinetics. J Biopharm Stat 2010; 20:
Anaesthesiol 2010; 23: 455–460. sedation. Emerg Med J 2004; 21: 271–272. 886–902.
2 Edrich T, Friedrich AD, Eltzschig HK et al. 15 Herd DW, Anderson BJ, Holford NH. Mod- 29 Foracchia M, Hooker A, Vicini P et al.
Ketamine for long-term sedation and analge- eling the norketamine metabolite in children POPED, a software for optimal experiment
sia of a burn patient. Anesth Analg 2004; 99: and the implications for analgesia. Pediatr design in population kinetics. Comput Meth-
893–895, table of contents. Anesth 2007; 17: 831–840. ods Programs Biomed 2004; 74: 29–46.
3 Chang LC, Raty SR, Ortiz J et al. The 16 Green SM, Hummel CB, Wittlake WA et al. 30 Tod M, Rocchisani JM. Implementation of
emerging use of ketamine for anesthesia and What is the optimal dose of intramuscular OSPOP, an algorithm for the estimation of
sedation in traumatic brain injuries. CNS ketamine for pediatric sedation? Acad Emerg optimal sampling times in pharmacokinetics
Neurosci Ther 2013; 19: 390–395. Med 1999; 6: 21–26. by the ED, EID and API criteria. Comput
4 Dallimore D, Herd DW, Short T et al. Dos- 17 Herd D, Anderson BJ. Ketamine disposition Methods Programs Biomed 1996; 50: 13–22.
ing ketamine for pediatric procedural seda- in children presenting for procedural seda- 31 Bazzoli C, Retout S, Mentre F. Design evalu-
tion in the emergency department. Pediatr tion and analgesia in a children’s emergency ation and optimisation in multiple response
Emerg Care 2008; 24: 529–533. department. Pediatr Anesth 2007; 17: 622– nonlinear mixed effect models: PFIM 3.0.
5 Grant IS, Nimmo WS, McNicol LR et al. 629. Comput Methods Programs Biomed 2010; 98:
Ketamine disposition in children and adults. 18 Kuhn E, Lavielle M. Maximum likelihood 55–65.
Br J Anaesth 1983; 55: 1107–1111. estimation in nonlinear mixed effects models. 32 D’Argenio DZ. Incorporating prior parame-
6 Lin C, Durieux ME. Ketamine and kids: an Comput Stat Data Anal 2005; 49: 1020–1038. ter uncertainty in the design of sampling
update. Pediatr Anesth 2005; 15: 91–97. 19 Stockmann C, Sherwin CM, Zobell JT et al. schedules for pharmacokinetic parameter
7 Herd D, Anderson B. Lack of pharmacoki- Population pharmacokinetics of intermittent estimation experiments. Math Biosci 1990;
netic information in children leads clinicians vancomycin in children with cystic fibrosis. 99: 105–118.
to use experience and trial-and-error to Pharmacotherapy 2013; 33: 1288–1296. 33 Morton NS. Ketamine for procedural seda-
determine how best to administer ketamine. 20 Holford NH. A size standard for pharmaco- tion and analgesia in pediatric emergency
Ann Emerg Med 2007; 49: 824, 824 e821; kinetics. Clin Pharmacokinet 1996; 30: 329– medicine: a UK perspective. Pediatr Anesth
author reply 825. 332. 2008; 18: 25–29.
8 Dachs RJ, Innes GM. Intravenous ketamine 21 Anderson BJ, Meakin GH. Scaling for 34 Ng K, Ang SY. Sedation with ketamine for
sedation of pediatric patients in the emer- size: some implications for paediatric paediatric procedures in the emergency
gency department. Ann Emerg Med 1997; 29: anaesthesia dosing. Paediatr Anaesth 2002; department – a review of 500 cases. Singa-
146–150. 12: 205–219. pore Med J 2002; 42: 300–304.
9 Green SM, Nakamura R, Johnson NE. Ke- 22 Akaike H. Information theory as an exten- 35 Green SM, Roback MG, Krauss B et al. Pre-
tamine sedation for pediatric procedures: sion of the maximum likelihood principle. In: dictors of airway and respiratory adverse
Part 1, A prospective series. Ann Emerg Med Petrov BN, Csaksi F, eds. Proceedings of the events with ketamine sedation in the emer-
1990; 19: 1024–1032. 2nd International Symposium on Informa- gency department: an individual-patient data
10 Green SM, Johnson NE. Ketamine sedation tion Theory. Budapest, Hungary: Akademiai meta-analysis of 8,282 children. Ann Emerg
for pediatric procedures: Part 2, Review and Kiado, 1973: 267–281. Med 2009; 54: 158–168, e151–e154.
implications. Ann Emerg Med 1990; 19: 23 Schwarz G. Estimating the dimension of a 36 Green SM, Roback MG, Krauss B et al. Pre-
1033–1046. model. Ann Stat 1978; 6: 461–464. dictors of emesis and recovery agitation with
11 Green SM, Rothrock SG, Lynch EL et al. 24 D’Argenio DZ, Schumitzky A, Wang X. emergency department ketamine sedation: an
Intramuscular ketamine for pediatric seda- ADAPT 5 User’s Guide: Pharmacokinetic/ individual-patient data meta-analysis of
tion in the emergency department: safety pro- Pharmacodynamic Systems Analysis Soft- 8,282 children. Ann Emerg Med 2009; 54:
file in 1,022 cases. Ann Emerg Med 1998; 31: ware. Los Angeles: Biomedical Simulations 171–180, e171–e174.
688–697. Resource, 2009. 37 Forrest ABC, Nix D, Schentag JJ. A new
12 McCarty EC, Mencio GA, Walker LA et al. 25 D’Argenio DZ. Optimal sampling times for approach for designing population sparse
Ketamine sedation for the reduction of chil- pharmacokinetic experiments. J Pharmacoki- sampling strategies – applied to ciprofloxacin
dren’s fractures in the emergency depart- net Biopharm 1981; 9: 739–756. pharmacokinetics. Clin Pharmacol Ther
ment. J Bone Joint Surg Am 2000; 82-A: 912– 26 Tam VH, Preston SL, Drusano GL. Optimal 1991; 49: 153.
918. sampling schedule design for populations of 38 Barrett JS, Della Casa Alberighi O, Laer S
13 Lawrence LM, Wright SW. Sedation of pedi- patients. Antimicrob Agents Chemother 2003; et al. Physiologically based pharmacokinetic
atric patients for minor laceration repair: 47: 2888–2891. (PBPK) modeling in children. Clin Pharma-
effect on length of emergency department 27 Sheiner LB, Beal SL. Some suggestions for col Ther 2012; 92: 40–49.
stay and patient charges. Pediatr Emerg Care measuring predictive performance. J Phar- 39 Reed MD. Optimal sampling theory: an
1998; 14: 393–395. macokinet Biopharm 1981; 9: 503–512. overview of its application to pharmacoki-
14 Green SM, Krauss B. Ketamine is a safe, 28 Foo LK, Duffull S. Methods of robust design netic studies in infants and children. Pediat-
effective, and appropriate technique for of nonlinear models with an application to rics 1999; 104: 627–632.

216 © 2014 John Wiley & Sons Ltd

Pediatric Anesthesia 25 (2015) 211–216
Copyright of Pediatric Anesthesia is the property of Wiley-Blackwell and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.