ORIGINAL www.jpeds.

com • THE JOURNAL OF PEDIATRICS

ARTICLES
Extubation Failure after Neonatal Cardiac Surgery: A Multicenter Analysis
Christopher W. Mastropietro, MD1, Katherine Cashen, DO2, Lisa M. Grimaldi, MD3, Keshava Murty Narayana Gowda, MBBS4,
Kurt D. Piggott, MD5, Michael Wilhelm, MD6, Eleanor Gradidge, MD7, Elizabeth A. S. Moser, MS8, Brian D. Benneyworth, MD1,
and John M. Costello, MD9

Objectives To describe the epidemiology of extubation failure and identify risk factors for its occurrence in a mul-
ticenter population of neonates undergoing surgery for congenital heart disease.
Study design We conducted a prospective observational study of neonates ≤30 days of age who underwent
cardiac surgery at 7 centers within the US in 2015. Extubation failure was defined as reintubation within 72 hours
of the first planned extubation. Risk factors were identified with the use of multivariable logistic regression analy-
sis and reported as OR with 95% CIs. Multivariable logistic regression analysis was conducted to examine the re-
lationship between extubation failure and worse clinical outcome, defined as hospital length of stay in the upper
25% or operative mortality.
Results We enrolled 283 neonates, of whom 35 (12%) failed their first extubation at a median time of 7.5 hours
(range 1-70 hours). In a multivariable model, use of uncuffed endotracheal tubes (OR 4.6; 95% CI 1.8-11.6) and
open sternotomy of 4 days or more (OR 4.8; 95% CI 1.3-17.1) were associated independently with extubation failure.
Accordingly, extubation failure was determined to be an independent risk factor for worse clinical outcome (OR
5.1; 95% CI 2-13).
Conclusions In this multicenter cohort of neonates who underwent surgery for congenital heart disease, extu-
bation failure occurred in 12% of cases and was associated independently with worse clinical outcome. Use of
uncuffed endotracheal tubes and prolonged open sternotomy were identified as independent and potentially modi-
fiable risk factors for the occurrence of this precarious complication. (J Pediatr 2017;182:190-6).

A
s surgical techniques and perioperative management of neonates undergoing surgery for congenital heart disease con-
tinue to evolve, mechanical ventilation continues to be a necessary and important component of postoperative
management.1,2 In a small but important number of neonates, initial attempts at extubation fail and reintubation is re-
quired. Reintubation of these fragile patients is a high-risk procedure with potential for life-threatening consequences such as
profound hypoxemia, cardiovascular instability, and cardiopulmonary arrest. Following reintubation, these neonates are then
committed to another course of mechanical ventilation, with its risks and exposures such as ventilator-associated infections,
airway trauma, and the need for sedative infusions. In neonates and children un-
dergoing cardiac surgery, extubation failure has been associated consistently with
increased postoperative morbidity and mortality.3-7 Prevention of this important
complication is therefore an essential part of optimizing clinical outcomes in this From the 1Department of Pediatrics, Division of Critical
patient population. Care, Indiana University School of Medicine, Riley
Hospital for Children, Indianapolis, IN; 2Division of Critical
Unfortunately, assessment of extubation readiness in critically ill neonates can Care, Department of Pediatrics, Wayne State University
be challenging.8 In many cases, despite reassuring respiratory and hemodynamic School of Medicine, Children’s Hospital of Michigan,
Detroit, MI; 3Division of Cardiovascular Intensive Care,
indices, extubation failure still occurs. Initial studies focused on extubation failure Phoenix Children’s Hospital, Department of Child Health,
University of Arizona, College of Medicine – Phoenix,
after surgery for congenital heart disease in neonates, although providing some Phoenix, AZ; 4Division of Pediatric Critical Care Medicine,
Department of Pediatrics, Cleveland Clinic, Cleveland,
valuable insight, have been restricted to small, single-center cohorts and thus have OH; 5The Heart Center at Arnold Palmer Hospital for
had limited generalizability.6,7,9 In a recent review of data from the Pediatric Cardiac Children, Division of Pediatric Cardiac Intensive Care,
Department of Pediatrics, University of Central Florida
Critical Care Consortium (PC4), only the presence of an underlying anatomic airway College of Medicine, Orlando, FL; 6Division of Pediatric
Cardiac Intensive Care, Department of Pediatrics,
anomaly could be identified as an independent risk factor for extubation failure University of Wisconsin, Madison, WI; 7Division of Critical
Care at Phoenix Children’s Hospital, Department of Child
after neonatal cardiac surgery, but the analysis in this study was limited to the vari- Health, University of Arizona, College of Medicine –
ables available within the registry.10 We aimed to describe the epidemiology of ex- Phoenix, Phoenix, AZ; 8Department of Biostatistics,
Indiana University School of Medicine & Richard M.
tubation failure in a large multicenter population of neonates undergoing surgery Fairbanks School of Public Health, Indianapolis, IN; and
9
Divisions of Cardiology & Critical Care Medicine,
Department of Pediatrics, Ann & Robert H. Lurie
Children’s Hospital of Chicago, Northwestern University
Feinberg School of Medicine, Chicago, IL
The authors declare no conflicts of interest.
ICU Intensive care unit Portions of the study were presented at the annual
LOS Length of stay meeting of the Pediatric Cardiac Intensive Care Society,
Houston, TX, December 10-11, 2015.
PC4 Pediatric Cardiac Critical Care Consortium
STAT Mortality Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery 0022-3476/$ - see front matter. © 2016 Elsevier Inc. All rights
Category Society Congenital Heart Surgery Mortality Category reserved.
http://dx.doi.org10.1016/j.jpeds.2016.12.028

190
 Volume 182 • March 2017

for congenital heart disease. From these data, we aimed to iden- logistic regression model. The multivariable model also was
tify risk factors for extubation failure and determine its impact analyzed as a mixed model with a random effect for center.
on clinical outcome. Linearity in the logit was examined for continuous variables
before model-building; those with evidence of nonlinearity were
Methods converted to categorical variables. Variables with P values < .05
after multivariable analysis were then identified as indepen-
All neonates who underwent surgery for congenital heart disease dent risk factors for extubation failure after neonatal cardiac
at 7 tertiary care pediatric referral centers between January 1 surgery.
and December 31, 2015, were considered for inclusion in the To determine whether extubation failure had an indepen-
study. Patients were enrolled prospectively from the follow- dent effect on clinical outcome, we dichotomized hospital length
ing institutions: Riley Hospital for Children, Indianapolis, of stay (LOS) as upper 25% and lower 75% and then defined
Indiana (coordinating center); Children’s Hospital of Michi- prolonged LOS as patients in the upper (worst) 25%. Pa-
gan, Detroit, Michigan; Cleveland Clinic, Cleveland, Ohio; tients with prolonged LOS and patients who died before the
Arnold Palmer Hospital for Children, Orlando, Florida; Phoenix cut-off value were categorized as having worse clinical out-
Children’s Hospital, Phoenix, Arizona; Ann & Robert H. comes compared with the rest of the cohort. Stepwise multi-
Lurie Children’s Hospital of Chicago, Illinois; and American variable regression modeling was then performed to determine
Family Children’s Hospital, Madison, Wisconsin. whether extubation failure was associated independently with
The study was approved by the institutional review boards worse clinical outcomes. All statistical analyses were per-
at all participating centers and was performed in accordance formed with Stata version 14 (StataCorp, College Station, Texas)
with the ethical standards laid down in the 1964 Declaration and SAS version 9.4 (SAS Institute, Cary, North Carolina).
of Helsinki and its later amendments. Because of the obser-
vational nature of the data collected, the need for informed Results
consent was waived. Neonates <2.5 kg who underwent iso-
lated ligation of patent ductus arteriosus, neonates with tra- We prospectively enrolled 293 neonates who underwent surgery
cheostomy tubes in place at the time of cardiovascular surgery, for congenital or acquired heart disease in 2015. Ten patients
and neonates who died or underwent tracheostomy without died before any extubation attempts and were therefore not
undergoing any planned extubation attempts were excluded included in the analysis. Data were collected for the remain-
from the analysis. ing 283 neonates, who were intubated endotracheally, re-
Postoperative management and the decision to extubate from ceived mechanical ventilation during their surgical repair or
mechanical ventilation was left to the discretion of the primary palliation, and had at least one extubation attempt. Median
intensive care unit (ICU) teams at each institution. Postop- number of subjects enrolled at each institution was 30 (range
erative management, assessment of extubation readiness, and 17-71). Four institutions were categorized as small-to-moderate
assessment of endotracheal tube leak were determined by the volume centers (range 17-30 neonates), and 3 institutions were
standard practices of individual study centers and the clini- categorized as large volume (range 55-71 neonates). Primary
cal judgment of individual clinicians working in those centers. surgical procedures are provided in Table II (available at
Extubation failure was defined as the need for reintubation www.jpeds.com), organized by Society of Thoracic Surgeons-
within 72 hours of first attempted extubation from mechani- European Association for Cardiothoracic Surgery Society Con-
cal ventilation.7 Patients who underwent endotracheal intu- genital Heart Surgery Mortality Category (STAT Mortality
bation and had another period of mechanical ventilation that Category).11,13 The most common operations performed were
was initiated after 72 hours from the first extubation attempt the arterial switch procedure (n = 60) and the Norwood pro-
were considered successful initial extubations. A comprehen- cedure (n = 47).
sive list of variables collected and additional definitions are in- Thirty-five patients (12%) failed their first extubation attempt
cluded in Table I (available at www.jpeds.com).11-19 at a median time of 7.5 hours after the attempt (range 1-70
hours). The timing of extubation failure is summarized in
Statistical Analyses Figure 1. Extubation failure ranged from 8% to 23% across
Data are represented with the use of descriptive statistics as the participating centers. Extubation failures were distrib-
follows: means with SDs for continuous normally distributed uted broadly across STAT Mortality Categories and surgical pro-
variables, medians with 25th percentiles and 75th percentiles cedures (Table II). The most common diagnosis implicated as
for continuous skewed variables, and absolute counts with per- the major contributor to extubation failure was cardiogenic
centages for categorical variables. To determine the risk factors shock (n = 11, 31%). Other clinical findings implicated as im-
associated with extubation failure, we performed a bivariate portant contributors to the extubation failures were pulmo-
analysis by comparing variables in neonates who required nary edema (n = 7), apnea/hypopnea (n = 5), atelectasis (n = 3),
reintubation within 72 hours with those who were extubated chylous effusion (n = 3), diaphragm paresis (n = 3), stridor
successfully on their first attempt using t tests, Mann-Whitney (n = 2), and hypercarbia (n = 1). Vocal cord paresis did not con-
U tests, c2 tests, and Fisher exact test as appropriate for indi- tribute to any of the extubation failures in this cohort. Eight
vidual variables. All variables with P values < .2 on bivariate patients (23%) had residual cardiac lesions that were deemed
analysis were considered for inclusion in our multivariate to have possibly contributed to their extubation failures:
191
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 182

illustrated in Figure 2, this increased prevalence of uncuffed

endotracheal tubes largely was responsible for the observed dif-
ference in endotracheal tube size. Postoperatively, patients who
failed extubation were more likely to have tachyarrhythmias
and had longer durations of mechanical ventilation before their
first extubation attempts. At the time of extubation, positive
end-expiratory pressure tended to be greater than 5 cmH2O
more frequently in patients who failed extubation. The use of
prophylactic periextubation dexamethasone was not statisti-
cally different between groups.
There also was a notable increase in the likelihood of ex-
tubation failure in patients who had an open sternotomy for
4 days or more. No patient in the study developed mediasti-
nitis, and sternal wound infections were observed in 5 pa-
Figure 1. Timing of extubation failures: Most extubation fail- tients, only 1 of whom had an open sternotomy for 4 or more
ures occurred within the first 12 hours after extubation. days. Patients who required open sternotomy for more than
4 days were, however, more likely to receive neuromuscular
blockade (87%) as compared with the rest of the cohort (38%,
P < .001).
3 patients with 2 ventricle physiology had residual aortic Two patients were extubated in the operating room imme-
coarctations, 2 patients had moderate-to-large ventricular septal diately after surgery, and the other 281 patients were extu-
defects, 1 patient had moderate mitral regurgitation, 1 patient bated postoperatively in the ICU. Respiratory support provided
with hypoplastic left heart syndrome had moderate tricuspid on extubation is compared in Table V, and the proportion of
regurgitation, and 1 patient had a small non–apex-forming left patients who required escalation of support are described
ventricle. graphically in Figure 3. Most neonates (n = 155, 55%) were
Two patients suffered cardiac arrests at the time of extubated to high-flow nasal cannula. The remaining neo-
reintubation. One of these patients was a male neonate born nates were extubated to room air or oxygen via nasal cannula
with hypoplastic left heart syndrome who underwent the (n = 87, 31%) and noninvasive positive pressure ventilation
Norwood procedure with Sano modification at 11 days of age (n = 41, 14%). High-flow cannula was the preferred ap-
and was extubated on postoperative day 12. He developed pro- proach, but no significant advantage was observed with this
gressively worsening cardiac failure and cardiogenic shock, and modality or noninvasive positive pressure in regards to extu-
on reintubation at 59 hours postextubation, he experienced bation failure.
cardiac arrest, prolonged cardiopulmonary resuscitation, and The results of our multivariable logistic regression analy-
emergent cannulation for extracorporeal membrane oxygen- sis are provided in Table VI. Use of an uncuffed endotra-
ation. Despite these efforts, the child ultimately died. The second cheal tube and duration of open sternotomy of 4 days or more
patient was a male neonate with Williams syndrome born with were associated independently with extubation failure, in a
valvar and supravalvar aortic stenosis and coarctation of the model that also included hypoplastic left heart syndrome, un-
aorta who underwent coarctectomy via end-to-end anasto- derlying anatomy airway or respiratory anomaly, positive end-
mosis at 3 days of age. He was extubated on postoperative day expiratory pressure greater than 5 at the time of extubation,
2 then acutely deteriorated at 26 hours postextubation result- and duration of mechanical ventilation. Additional variables
ing in cardiac arrest. He was reintubated immediately, but re- that were significant on bivariate analysis such as the pres-
suscitation efforts were unsuccessful and the patient did not ence of underlying genetic anomalies, extracorporeal mem-
survive. brane oxygenation, postoperative tachyarrhythmias, and serum
Preoperative, perioperative, and postoperative characteris- lactate were not significant on multivariable analysis and had
tics of patients who failed their first extubation attempt are little effect on the multivariable model. Moreover, the effect
compared with patients who were extubated successfully in of use of uncuffed endotracheal tubes and open sternotomy
Tables III-V (available at www.jpeds.com). Preoperatively, pa- for 4 or more days remained significant when other markers
tients who failed extubation were more likely to have hypo- of disease complexity and severity such as STAT Mortality Cat-
plastic left heart syndrome or an underlying airway anomaly egory, duration of cardiopulmonary bypass, use of inhaled nitric
or to require endotracheal intubation and remain on me- oxide, vasoactive inotrope score at 48 hours, or presence of a
chanical ventilation until surgery. Center volume or the pres- clinical significant residual lesion were added to the model. In
ence of a dedicated cardiac ICU did not appear to influence the multivariable mixed model, the random effect of center
the likelihood of extubation failure. Perioperatively, mean en- was estimated to be zero.
dotracheal tube inner diameter size (indexed to body surface Clinical outcomes were significantly worse in patients who
area) was greater in patients who failed extubation. In addi- failed extubation (Table VII; available at www.jpeds.com). Neo-
tion, neonates who failed extubation were more likely to have nates who failed their first extubation had significantly longer
been intubated with uncuffed endotracheal tubes and, as ICU and hospital stays as compared with patients who were
192 Mastropietro et al
March 2017 ORIGINAL ARTICLES

Figure 2. Analysis of endotracheal (ET) tube size with: mean ET tube inner diameter size was greater in neonates who failed
extubation as compared with neonates who extubated successfully, P = .014. Neonates with uncuffed tubes who failed extu-
bation (n = 12) largely were responsible for this observed difference, as the mean inner diameter of their ET tubes was sub-
stantially greater than that of the neonates with cuffed endotracheal tubes who failed extubation (n = 23), P = .008. Data represented
as mean with SE. *Indicates statistical significance.

extubated successfully, and mortality was significantly greater clinical outcome included single ventricle anatomy, genetic or
in neonates who failed their first extubation, 23% as com- chromosomal abnormality, presence of any noncardiac ana-
pared with 0.4%. Neonates with lengths of hospital stay in the tomic abnormality, occurrence of infection or necrotizing en-
upper 25% (>36 days) and neonates who died were catego- terocolitis during hospitalization (ie, pre- or postoperatively),
rized as having worse clinical outcomes. After stepwise mul- presence of a clinically significant residual lesion, and dura-
tivariable regression analysis, extubation failure was an tion of mechanical ventilation before the first extubation
independent predictor of worse outcome after neonatal car- attempt. The results of this multivariable logistic model are pro-
diovascular surgery. Other independent predictors of worse vided in Table VIII (available at www.jpeds.com).

Figure 3. Respiratory support provided on extubation: number of neonates extubated to room air or oxygen via nasal cannula
(NC), oxygen via high-flow nasal cannula (HFNC), and noninvasive positive pressure ventilation (NIPPV). Each bar represents
neonates successfully extubated (green), neonates in which support was escalated to HFNC or NIPPV but not reintubated (yellow),
and neonates who required reintubation (red).

Extubation Failure after Neonatal Cardiac Surgery: A Multicenter Analysis 193

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 182

with uncuffed tubes, and these tubes remained in use until the
Table VI. Multivariate logistic regression analysis for pre-
first postoperative extubation attempt. Based on our data, phy-
dictors of extubation failure
sicians in our study chose endotracheal tubes with signifi-
Variables OR 95% CI P value cantly larger inner diameters (indexed to body surface area)
Hypoplastic left heart 1.9 0.7-4.7 .19 when using uncuffed tubes, presumably to ensure adequate ven-
Underlying airway anomaly 2.7 0.6-11.6 .18 tilation without an excessive leak. Some authors have as-
Uncuffed endotracheal tube 4.6 1.8-11.6 .001
Prolonged open sternotomy* 4.8 1.3-17.1 .02 serted that when these larger diameter uncuffed endotracheal
PEEP greater than 5 cmH2O† 2.7 0.99-7.3 .05 tubes are used, irritation of glottic and subglottic areas likely
Duration of mechanical ventilation‡ 1.3 0.3-5.9 .23 occurs, especially to the rigid cricoid ring and vocal folds, which
PEEP, positive end-expiratory pressure. are particularly susceptible to damage from mucosal shear.21
*Greater than or equal to 4 d. Stridor was only noted in 2 patients who failed extubation, but
†At the time of extubation.
‡Variable was nonlinear in the logit and was categorized into quintiles; OR represents longest significant injury to the upper airway is not always accompa-
quintile compared with shortest quintile. nied by stridor.24 Indeed, in neonates, small decreases in airway
diameter from airway edema or inflammation may not cause
overt stridor but could markedly increase airway resistance,
Discussion which is proportional to radius to the fourth power.8 In neo-
nates with underlying cardiac disease, this increase in airway
Extubation failure occurred in 12% of the patients in our het- resistance could overwhelm their limited cardiopulmonary
erogeneous population of neonates who underwent surgery reserve, increase ventricular wall stress (ie, afterload), and
for congenital heart disease. Our study is consistent with the promote extubation failure. Standardized use of smaller di-
recent review of the PC4 registry, which reported an extuba- ameter low-pressure cuffed endotracheal tubes may be a rela-
tion failure rate of 11% in 899 neonates who underwent cardiac tively simple way of decreasing the occurrence rate of extubation
surgery.10 Not surprisingly, the risk of extubation failure in neo- failure in neonates recovering from cardiac surgery.
nates recovering from cardiac surgery appears to be greater than The second potential modifiable risk factor identified herein
the general population of children receiving mechanical ven- is prolonged duration of open sternotomy. Although neo-
tilation in cardiac intensive care units, which recently was re- nates with prolonged open sternotomy likely had greater disease
ported to be 6% in a multicenter cohort.5 We observed a range burden in the early postoperative period, this variable re-
of extubation failure rates across the 7 participating centers mained significant after adjustment for many markers of disease
in our study, from 8% to 23%, also consistent with the data complexity and severity, including the presence of hypoplas-
from the PC4 registry, which reported a range of 5%-22%.10 tic left heart syndrome, STAT Mortality Category, duration of
Furthermore, the only retrospective single-center study that cardiopulmonary bypass, postoperative vasoactive inotrope
focused on a heterogeneous population of neonates who un- score, and duration of mechanical ventilation. Neuromuscu-
derwent cardiac surgery published an extubation failure rate lar blockade has been shown to potentiate the diaphragmatic
of 17.5%.13 This variation in extubation failure occurrence rates atrophy that can occur during mechanical ventilation.25 Al-
across centers suggests an opportunity for institutional quality though there was no difference in the number of patients who
assurance initiatives aimed at decreasing the occurrence of this received neuromuscular blockade in patients who failed ex-
frustrating and potentially life-threatening complication. tubation and patients who extubated successfully, we did not
As targets of these quality assurance initiatives, we identi- quantitate the cumulative dose or number of days that neu-
fied 2 potentially modifiable risk factors, the first of which was romuscular blockade was administered. Because nearly 90%
the use of uncuffed endotracheal tubes. Uncuffed endotra- of patients with prolonged open sternotomy received neuro-
cheal tubes historically have been used more commonly in pe- muscular blockade, we speculate that cumulative exposure to
diatric patients because of the notion that the narrowest point neuromuscular blockade was likely greater in these neonates
of the airway in young children is at the level of the circum- as compared with the rest of the cohort. If so, these patients
ferential, nondistensible cricoid cartilage below the vocal may have had more diaphragmatic weakness at the time of ex-
cords.20,21 Uncuffed endotracheal tubes, when inserted into these tubation. Other unknown effects of prolonged open ster-
airways, should optimally be large enough to seal the cricoid notomy on respiratory system function also could have played
ring but small enough to allow an air leak when airway pres- a role. Regardless of the pathophysiology behind our finding,
sures of 20-30 cmH2O are applied.22 As our understanding of earlier sternal closure when possible may improve the chances
the developmental changes of the upper airway has improved22 of successful neonatal extubation.
and modern low-pressure cuffed endotracheal tubes have Our data also suggest that the length of time mechanically
become readily available,23 uncuffed endotracheal tubes have ventilated does not independently affect the likelihood of ex-
fallen out of favor in some pediatric institutions. tubation success or failure in neonatal cardiac surgical pa-
Despite this evolution, use of uncuffed endotracheal tubes tients. Likewise, in the only previous single-center study to
still is relatively common for neonatal populations, where examine extubation failure in a heterogeneous group of neo-
concern for airway damage from excessive cuff pressures es- nates who underwent cardiac surgery and in the study re-
pecially is fervent. In fact, several of the neonates in our study cently presented from the PC4 database, no independent effect
were intubated endotracheally preoperatively by neonatologists of duration of mechanical ventilation on the likelihood of
194 Mastropietro et al
March 2017 ORIGINAL ARTICLES

extubation failure was found7,10 In contrast, recent data from across the 3 modalities should not be overinterpreted. Ran-
a large cohort of children of all age groups with critical cardiac domized controlled trials in this patient population would be
disease (the majority of which underwent cardiac surgery) dem- needed to determine whether there is a clear benefit or det-
onstrated duration of mechanical ventilation to be the most riment to the routine use of one of these therapies and should
important risk factor for extubation failure. 5 Hence, al- be pursued. Lastly, although we did not detect any center-
though most clinicians commonly have heightened aware- level effects on our findings, we acknowledge that there may
ness of the risk of extubation failure in patients who have be some intangible or immeasurable center-specific factors that
required prolonged mechanical ventilation, this increased could have influenced the likelihood of extubation success or
concern for extubation failure should be applied to all neo- failure.
nates following cardiac surgery, whether extubation is occur- In conclusion, our multicenter analysis has demonstrated
ring early or late in their course. an extubation failure rate of 12% in neonates who under-
We also found extubation failure to be associated indepen- went surgery for congenital heart disease. Efforts to reduce the
dently with worse outcome, in a multivariable model that in- occurrence of extubation failure could ultimately improve the
cluded several additional comorbidities and markers of disease clinical outcomes of these complex neonates. ■
severity. We do not assert that the extubation failures in and
of themselves were responsible for the worse outcomes but Submitted for publication Aug 18, 2016; last revision received Oct 18, 2016;
rather one of several possible contributors. Additional studies accepted Dec 8, 2016

on promising adjuncts to traditional extubation readiness as- Reprint requests: Christopher W. Mastropietro, MD, Department of Pediatrics,
Division of Critical Care, Indiana University School of Medicine, Riley Hospital
sessment protocols such as near-infrared spectroscopy for Children, 705 Riley Hospital Dr, Riley Phase 2, RI4119C, Indianapolis, IN
monitoring26 and dead space-to-tidal volume calculations27 46202. E-mail: [email protected]
should be pursued.
Our study has some limitations. Although this study in- References
cluded centers representing a reasonable mix of small-,
1. Polito A, Patorno E, Costello JM, Salvin JW, Emani SM, Rajagopal S, et al.
moderate-, and large-volume centers in different geographic
Perioperative factors associated with prolonged mechanical ventilation
locations, only 7 centers were involved and the number of ex- after complex congenital heart surgery. Pediatr Crit Care Med 2011;12:e122-
tubation failures in the study was not high. A study with a 6.
greater number of extubation failures potentially could have 2. Pagowska-Klimek L, Pychynska-Pokorska M, Krajewski W, Moll JJ. Pre-
power to identify other important risk factors. For example, dictors of long intensive care unit stay following cardiac surgery in chil-
dren. Eur J Cardiothorac Surg 2011;40:179-84.
unlike the review of the PC4 registry,10 underlying airway or
3. Baisch SD, Wheeler WB, Kurachek SC, Cornfield DN. Extubation failure
respiratory anomaly, which was only present in 13 of our pa- in pediatric intensive care incidence and outcomes. Pediatr Crit Care Med
tients, could not be determined to be an independent risk factor 2005;6:312-8.
for extubation failure in our study. In addition, although data 4. Harkel AD, van der Vorst MM, Hazekamp MG, Ottenkamp J. High mor-
for an extensive list of variables were collected, there were likely tality rate after extubation failure after pediatric cardiac surgery. Pediatr
Cardiol 2005;26:756-61.
some patient characteristics that could have contributed to ex-
5. Gaies M, Tabbutt S, Schwartz SM, Bird GL, Alten JA, Shekerdemian LS,
tubation failures but could not be analyzed in this observa- et al. Clinical epidemiology of extubation failure in the pediatric cardiac
tional study. For instance, across institutions, the level of ICU: a report from the pediatric cardiac critical care consortium. Pediatr
consciousness or nutritional state at the time of extubation were Crit Care Med 2015;16:837-45.
not be measured objectively, nor was the occurrence of nar- 6. Gupta P, McDonald R, Gossett JM, Butt W, Shinkawa T, Imamura M,
et al. A single-center experience of extubation failure in infants
cotic or sedative withdrawal consistently assessed as standard
undergoing the Norwood operation. Ann Thorac Surg 2012;94:1262-
practice, preventing us from determining whether these factors 8.
could have had a role in any of the extubation failures. The 7. Laudato N, Gupta P, Walters HL 3rd, Delius RE, Mastropietro CW. Risk
assessment of extubation readiness was not protocolized, factors for extubation failure following neonatal cardiac surgery. Pediatr
because we did not aim to determine the effectiveness or in- Crit Care Med 2015;16:859-67.
8. Newth CJ, Venkataraman S, Willson DF, Meert KL, Harrison R, Dean JM,
effectiveness of different extubation readiness strategies with
et al. Weaning and extubation readiness in pediatric patients. Pediatr Crit
this dataset. Prospective studies examining specific extuba- Care Med 2009;10:1-11.
tion readiness methods (eg, trials of continuous positive airway 9. Gupta P, McDonald R, Goyal S, Gossett JM, Imamura M, Agarwal A, et al.
pressure with pressure support) are needed to appropriately Extubation failure in infants with shunt-dependent pulmonary blood flow
address this question. We also were unable to gain additional and univentricular physiology. Cardiol Young 2014;24:64-72.
10. Mastropietro CW, Benneyworth B, Graham E, Klugman D, Costello JM,
insight into the aforementioned adjuncts to extubation readi-
Zhang W, et al. Multi-institutional analysis of extubation failure after neo-
ness assessment (eg, near-infrared spectroscopy and dead space- natal cardiac surgery. Crit Care Med 2016;43:19.
to tidal-volume ratio), because these modalities were 11. STS Congenital Heart Surgery Database v3.3. http://www.sts.org/node/
implemented inconsistently at the participating institutions. 42208. Accessed April 1, 2016.
Furthermore, our study was not designed to detect advan- 12. Pasquali SK, Hall M, Li JS, Peterson ED, Jaggers J, Lodge AJ, et al. Cor-
ticosteroids and outcome in children undergoing congenital heart surgery:
tages or disadvantages to the various modalities available for
analysis of the Pediatric Health Information Systems database. Circula-
respiratory support upon extubation (eg, oxygen via nasal tion 2010;122:2123-30.
cannula, high-flow nasal cannula, or noninvasive positive pres- 13. Jacobs JP, O’Brien SM, Pasquali SK, Jacobs ML, Lacour-Gayet FG,
sure), and thus the observed differences in extubation failure Tchervenkov CI, et al. Variation in outcomes for risk-stratified pediatric

Extubation Failure after Neonatal Cardiac Surgery: A Multicenter Analysis 195

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 182

cardiac surgical operations: an analysis of the STS Congenital Heart Surgery 20. Taylor C, Subaiya L, Corsino D. Pediatric cuffed endotracheal tubes: an
Database. Ann Thorac Surg 2012;94:564-71. evolution of care. Ochsner J 2011;11:52-6.
14. Butts RJ, Scheurer MA, Atz AM, Zyblewski SC, Hulsey TC, Bradley SM, 21. Litman RS, Maxwell LG. Cuffed versus uncuffed endotracheal tubes in
et al. Comparison of maximum vasoactive inotropic score and low cardiac pediatric anesthesia: the debate should finally end. Anesthesiology
output syndrome as markers of early postoperative outcomes after neo- 2013;118:500-1.
natal cardiac surgery. Pediatr Cardiol 2012;33:633-8. 22. Litman RS, Weissend EE, Shibata D, Westesson PL. Developmental changes
15. Gaies MG, Jeffries HE, Niebler RA, Pasquali SK, Donohue JE, Yu S, et al. of laryngeal dimensions in unparalyzed, sedated children. Anesthesiol-
Vasoactive-inotropic score is associated with outcome after infant cardiac ogy 2003;98:41-5.
surgery: an analysis from the Pediatric Cardiac Critical Care Consor- 23. Newth CJ, Rachman B, Patel N, Hammer J. The use of cuffed versus
tium and Virtual PICU System Registries. Pediatr Crit Care Med uncuffed endotracheal tubes in pediatric intensive care. J Pediatr
2014;15:529-37. 2004;144:333-7.
16. Davidson J, Tong S, Hancock H, Hauck A, da Cruz E, Kaufman J. Pro- 24. Holzki J, Laschat M, Puder C. Stridor is not a scientifically valid outcome
spective validation of the vasoactive-inotropic score and correlation to measure for assessing airway injury. Paediatr Anaesth 2009;19(suppl 1):180-
short-term outcomes in neonates and infants after cardiothoracic surgery. 97.
Intensive Care Med 2012;38:1184-90. 25. Testelmans D, Maes K, Wouters P, Gosselin N, Deruisseau K, Powers S,
17. Miletic KG, Delius RE, Walters HL 3rd, Mastropietro CW. Prospective vali- et al. Rocuronium exacerbates mechanical ventilation-induced
dation of a novel vasoactive-ventilation-renal score as a predictor of out- diaphragm dysfunction in rats. Crit Care Med 2006;34:3018-
comes after pediatric cardiac surgery. Ann Thorac Surg 2016;101:1558- 23.
63. 26. Foster CB, Spaeder MC, McCarter RJ, Cheng YI, Berger JT. The use of
18. Paret G, Ziv T, Barzilai A, Ben-Abraham R, Vardi A, Manisterski Y, et al. near-infrared spectroscopy during and extubation readiness trial as a pre-
Ventilation index and outcome in children with acute respiratory dis- dictor of extubation outcome. Pediatr Crit Care Med 2013;14:587-
tress syndrome. Pediatr Pulmonol 1998;26:125-8. 92.
19. Akcan-Arikan A, Zappitelli M, Loftis LL, Washburn KK, Jefferson LS, Gold- 27. Hubble CL, Gentile MA, Tripp DS, Craig DM, Meliones JN, Cheifetz IM.
stein SL. Modified RIFLE criteria in critically ill children with acute kidney Deadspace to tidal volume ratio predicts successful extubation in infants
injury. Kidney Int 2007;71:1028-35. and children. Crit Care Med 2000;28:2034-40.

196 Mastropietro et al
March 2017 ORIGINAL ARTICLES

Table I. Data collection and definitions

Preoperative data
• Age
• Sex
• Anthropometric data
• Estimated gestational age at birth
• Prematurity, ie, <37 weeks of gestation
• Underlying cardiovascular diagnoses
• Presence of genetic anomalies or noncardiac anatomic anomalies
• Need for preoperative mechanical ventilation
• Preoperative shock, defined as pH less than 7.2 or lactate greater than 4 mg/dL, to be consistent with the definition used by Society of Thoracic
Surgeons-Congenital Heart Surgery Database11
• Infection, ie, clinically relevant positive blood, urine, respiratory, or wound cultures, or culture-negative sepsis (ie, systemic inflammatory response syndrome with
suspected infection) treated with ≥7 days of antimicrobial therapy12
• Necrotizing enterocolitis
• Stroke or seizure, ie, patients with clinical or subclinical seizure activity were recorded as having seizures
• Whether patient was cared for in a dedicated cardiac ICU or multidisciplinary ICU
• Center size, ie, centers were arbitrarily categorized as small-to-moderate volume (<50 neonatal surgeries per year) and large volume (>50 neonatal surgeries per
year)

Perioperative data
• Use of systemic corticosteroids
• Endotracheal tube size and presence or absence of a cuff
• STAT Mortality Category11,13

Perioperative data
• Duration of cardiopulmonary bypass and aortic cross clamping
• Use of deep hypothermic arrest or antegrade cerebral perfusion

Postoperative data
• Admission and pre-extubation vasoactive medication requirements
o Vasoactive medication requirements were quantified using the VIS,14-17 calculated according to following formula: VIS = dopamine (mg/kg/min) + dobutamine
(mg/kg/min) + 100 × epinephrine (mg/kg/min) + 10 × milrinone (mg/kg/min) + 10 000 × vasopressin (U/kg/min) + 100 × norepinephrine (mg/kg/min)
• Admission and pre-extubation ventilator settings and arterial blood gas measurements
o Ventilation support at the time of admission was quantitated using the ventilation index,18 which was calculated according to the following formula:
VI = RR × (PIP – PEEP) × PaCO2/1000
• Use of extracorporeal membrane oxygenation
• Use of inhaled nitric oxide therapy
• Use of cardiopulmonary resuscitation
• Presence and duration of open sternotomy
• Maximum doses of fentanyl, morphine, midazolam, or dexmedetomidine infusions, if used
o Narcotic doses are reported in morphine equivalents
o Fentanyl doses were converted to morphine equivalents by multiplying the doses by 80, based on standard conversion tables
• Use of neuromuscular blockade
• Occurrence of postoperative arrhythmia
• Acute kidney injury defined by pRIFLE criteria19

Postoperative data
• Infection or necrotizing enterocolitis
• Stroke or seizure
• Use of systemic corticosteroids
• Fluid balance from ICU arrival to the first extubation attempt
• Presence of clinically important residual cardiac lesions, ie, residual valvular regurgitation described to be trivial or mild and residual septal defects described to be
small on postoperative echocardiograms were not included as clinically relevant residual cardiac lesions
• Respiratory support provided on extubation and whether or not the support was escalated
• Timing and presumed reason for reintubation, if it occurred

Outcome data
• Duration of mechanical ventilation before the first extubation attempt
• Total duration of mechanical ventilation
• Duration of intensive care unit
• Hospital LOS
• In-hospital mortality

PEEP, positive end-expiratory pressure; PIP, peak inspiratory pressure; pRIFLE, pediatric Risk, Injury, Failure, Loss, End-Stage Renal Disease; RR, respiratory rate; VI, ventilation index; VIS, vaso-
active inotrope score.

Extubation Failure after Neonatal Cardiac Surgery: A Multicenter Analysis 196.e1

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 182

Table II. Cardiovascular surgical procedures orga- Table III. Demographic, anthropometric, and preopera-
nized by STAT Mortality Categories tive data
Extubation Extubation All Extubation Extubation
All patients success failure patients success failure
Primary surgical procedures (N = 283) (n = 248) (n = 35) Variables (N = 283) (n = 248) (n = 35) P value
STAT Mortality Category 1 or 2 50 (18%) 40 (16%) 10 (29%) Gestational age at 38.5 ± 1.7 38.6 ± 1.7 38.1 ± 1.8 .08
CoA, end-to-end anastomosis 38 33 5 birth, wk
Repair, RVOT obstruction 7 5 2 Age at surgery, d 9±6 9±6 9±5 .56
Aortic valvotomy 2 0 2 Weight, kg 3.2 ± 0.6 3.3 ± 0.6 3.1 ± 0.7 .10
Other 2 2 1 Body surface area, m2 0.21 ± 0.10 0.22 ± 0.10 0.2 ± 0.03 .12
STAT Mortality Category 3 39 (14%) 37 (15%) 2 (6%) Sex, male 179 (63%) 158 (64%) 21 (60%) .67
Arterial switch, d-TGA with IVS 34 33 1 Prematurity, less than 39 (14%) 32 (13%) 7 (20%) .29
Patch aortoplasty 4 4 0 37 wk, n
Repair, CoA with ventricular 1 0 1 Genetic anomaly, n 41 (14%) 32 (13%) 9 (26%) .044
septal defect Down syndrome 5 3 2
STAT Mortality Category 4 140 (49%) 127 (51%) 13 (37%) Turner syndrome 3 3 0
Arterial switch, complex d-TGA* 26 25 1 DiGeorge syndrome 4 3 1
Systemic-to-PA shunt 33 29 4 Williams syndrome 2 1 1
Single ventricle 17 14 2 Other 27 22 5
Two ventricles 16 14 2 Any noncardiac anatomic 37 (13%) 29 (12%) 8 (23%) .10
PA banding 16 14 2 anomalies
Single ventricle 13 12 1 Anatomic airway/ 13 (5%) 9 (4%) 4 (11%) .06
Two ventricles 3 2 1 respiratory anomaly
PA banding + arch repair 9 7 2 Neurologic anomaly or 19 (7%) 15 (6%) 4 (11%) .27
Repair, TAPVC 17 17 0 insult*
Aortic arch reconstruction 15 14 1 Mechanical ventilation, n 130 (46%) 110 (44%) 20 (57%) .16
Repair, truncus arteriosus 9 9 0 Mechanical ventilation, d 0 (0, 2) 0 (0, 1) 1 (0, 2) .16
Repair, interrupted aortic arch 7 5 2 Mechanical ventilation to 70 (25%) 57 (23%) 13 (37%) .07
Other 8 7 1 surgery, n†
STAT Mortality Category 5 54 (19%) 44 (18%) 10 (29%) Shock, n 42 (15%) 36 (15%) 6 (17%) .68
Norwood procedure 46 37 9 Infection, n 14 (5%) 13 (5%) 1 (3%) 1.00
Blalock-Taussig shunt 16 14 2 Necrotizing enterocolitis, n 7 (2%) 6 (2%) 1 (3%) 1.00
Sano shunt 30 23 7 Single ventricle anatomy, n 93 (33%) 76 (31%) 17 (49%) .035
Hybrid procedure, HLHS 3 3 0 Hypoplastic left heart, n 59 (21%) 46 (19%) 13 (37%) .011
Damus-Kaye-Stanzel 2 1 1 Hypoplastic right heart, n 24 (8%) 22 (9%) 2 (6%) .75
Repair, TAPVC + systemic-to-PA 2 2 0 Heterotaxy, n 10 (4%) 8 (3%) 2 (6%) .36
shunt Dedicated cardiac ICU, n 214 (76%) 191 (77%) 23 (66%) .15
Yasui procedure 1 1 0 Large-volume center, n‡ 187 (66%) 166 (67%) 21 (60%) .42

CoA, coarctation of the aorta; d-TGA, d-transposition of the great arteries; HLHS, hypoplastic Continuous variables represented as mean (SD) or median (25th%, 75th%); categorical data
left heart syndrome; IVS, Intact Ventricular Septum; PA, pulmonary artery; RVOT, right ventricu- represented as absolute counts (%).
lar outflow tract obstruction; TAPVC, total anomalous pulmonary venous connections. *Underlying neuroanatomic anomaly, or stroke or seizures before surgery.
*d-TGA with ventricular septal defect or aortic coarctation. †After endotracheal intubation, remained on mechanical ventilation until surgery.
‡Greater than 50 neonatal cardiac surgeries in 2015.

196.e2 Mastropietro et al
March 2017 ORIGINAL ARTICLES

Table IV. Perioperative and ICU admission data

All patients Extubation success Extubation failure
Variables (N = 283) (n = 248) (n = 35) P value
Endotracheal tube characteristics
Diameter, mm/m2 15.4 ± 2.2 15.3 ± 2.2 16.3 ± 2.2 .014
Uncuffed, n 49 (17%) 37 (15%) 12 (34%) .005
Operative data
STAT Mortality Category 4 or 5, n 190 (67%) 168 (68%) 22 (63%) .57
Cardiopulmonary bypass, n 219 (77%) 194 (78%) 25 (71%) .37
Cardiopulmonary bypass, min 132 (51, 189) 132 (60, 193) 134 (0, 170) .48
Aortic cross clamp, min 52 (16, 91) 51 (18, 93) 57 (0, 78) .36
DHCA, n 68 (24%) 58 (23%) 10 (29%) .50
ACP, n 83 (29%) 69 (28%) 14 (40%) .14
DHCA and ACP, n 36 (13%) 28 (11%) 8 (23%) .10
Corticosteroids, n 211 (75%) 187 (66%) 24 (69%) .39
Delayed sternal closure, n 94 (33%) 78 (31%) 16 (46%) .09
Delayed sternal closure, d 0 (0, 2) 0 (0, 1) 0 (0, 4) .015
Delayed sternal closure ≥4 d, n 31 (11%) 20 (8%) 11 (31%) <.001
ECMO, n 15 (5%) 10 (4%) 5 (14%) .03
Clinically relevant residual lesion, n 48 (17%) 41 (17%) 7 (20%) .61
Postoperative admission data
pH 7.35 ± 0.08 7.35 ± 0.08 7.33 ± 0.07 .09
Lactate, mg/dL 3.7 ± 2.6 3.6 ± 2.4 4.7 ± 3.5 .021
Peak inspiratory pressure, mm Hg 22 ± 5 22 ± 5 23 ± 5 .11
PEEP, mm Hg 5±1 5±1 6±1 .10
Dynamic compliance, mL/cmH2O/kg 0.6 ± 0.2 0.6 ± 0.2 0.5 ± 0.2 .67
VI 19 (14, 25) 18 (14, 25) 21.1 (16, 26) .26
VIS 8 (5, 11) 8 (5, 11) 8 (0, 13) .51

ACP, antegrade cerebral perfusion; DHCA, deep hypothermic circulatory support; ECMO, extracorporeal membrane oxygenation.
Continuous variables represented as mean (SD) or median (25th%, 75th%); categorical data represented as absolute counts (%).

Extubation Failure after Neonatal Cardiac Surgery: A Multicenter Analysis 196.e3

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 182

Table V. Postoperative and pre-extubation data

All patients Extubation success Extubation failure
Variables (N = 283) (n = 248) (n = 35) P value
Postoperative data
Intra/postoperative CPR, n 12 (4%) 9 (4%) 3 (9%) .17
Inhaled nitric oxide use, n 64 (23%) 53 (21%) 11 (29%) .18
Any arrhythmia, n 73 (26%) 60 (24%) 13 (37%) .10
Tachyarrhythmias, n 54 (19%) 43 (17%) 11 (29%) .047
Necrotizing enterocolitis, n 6 (2%) 4 (2%) 2 (6%) .16
Infection, n 24 (8%) 21 (8%) 3 (9%) 1.00
Acute kidney injury, I or F, n* 57 (20%) 48 (19%) 9 (26%) .38
New seizures or stroke, n 11 (4%) 10 (4%) 1 (3%) 1.00
Fluid balance, mL/kg +4 (−51,+48) +5 (−49,+46) +1 (−63,+54) .85
Corticosteroid use
Precardiopulmonary bypass, n†
Hydrocortisone therapy, n‡ 91 (31%) 78 (31%) 13 (37%) .50
Periextubation dexamethasone, n§ 76 (27%) 64 (26%) 12 (34%) .29
Pain and sedation management
Narcotic infusion, n 228 (81%) 202 (81%) 26 (74%) .32
Max narcotic dose, mg/kg/h¶ 80 (25, 160) 80 (25, 160) 40 (20, 160) .22
Midazolam infusion, n 98 (35%) 86 (35%) 12 (34%) .96
Dexmedetomidine infusion, n 116 (41%) 102 (41%) 16 (46%) .90
Neuromuscular blockade, n 122 (43%) 104 (42%) 18 (54%) .29
Pre-extubation data
Pressure support, cmH2O 10 (10, 10) 10 (10, 10) 10 (10, 10) .37
Pressure support >10 cmH2O 40 (14%) 36 (15%) 4 (11%) .50
PEEP, cmH2O 5±1 5±1 5±1 .36
PEEP > 5 cmH2O 37 (13%) 29 (12%) 8 (23%) .07
VIS 5 (3, 6) 5 (3, 6) 5 (3, 7) .40
Mechanical ventilation, h** 72 (42, 134) 70 (42, 116) 138 (57, 238) <.001
Respiratory support provided on admission
Oxygen via nasal cannula 87 (31%) 81 (33%) 6 (17%) .06
Oxygen via high-flow nasal cannula 156 (55%) 131 (53%) 25 (71%) .038
Noninvasive positive pressure 40 (14%) 36 (15%) 4 (11%) .80

CPR, cardiopulmonary resuscitation.

Continuous variables represented as mean (SD) or median (25th%, 75th%); categorical data represented as absolute counts (%).
*Acute kidney injury categorized as “Injury” or “Failure” based on pRIFLE criteria.
†Intravenous methylprednisolone 30 mg/kg/dose or dexamethasone 10 mg/kg/dose.
‡Hydrocortisone provided for postoperative hemodynamic instability, 1 mg/kg/dose.
§Dexamethasone provided to peri-extubation to limit airway edema, 0.5 mg/kg/dose.
¶Provided in morphine equivalents.
**From ICU arrival after surgery to the first extubation attempt.

Table VIII. Multivariate logistic regression analysis for

predictors of worse clinical outcomes*
Table VII. Clinical outcomes Variables OR 95% CI P value
Extubation Extubation Single-ventricle anatomy 3.4 1.6-7 .001
All patients success failure Genetic or chromosomal abnormality 2.9 1.1-7.5 .03
Outcomes (N = 283) (n = 248) (n = 35) P value Noncardiac anatomic abnormality 3.7 1.7-9.6 .01
Clinically significant residual cardiac lesion 2.5 1.03-6.3 .04
Total duration of 83 (44, 178) 71 (42, 134) 359 (208, 749) <.001 Infection or necrotizing enterocolitis† 3 1.2-7.6 .02
ventilation, h Duration of mechanical ventilation‡ 1.01 1.005-1.013 <.001
ICU stay, d 11 (5, 20) 9 (5, 17) 30 (16, 61) <.001 Extubation failure 5.1 2-13 .001
Hospital stay, d 19 (11, 37) 17 (11, 30) 46 (26, 76) <.001
Mortality, n 9 (3.2%) 1 (0.4%) 8 (22.9%) <.001 *Defined as neonates with lengths of hospital stay in the upper 25% or death before hospital
discharge.
Continuous variables represented as median (25th%, 75th%); categorical data represented as †Preoperatively or postoperatively before first extubation attempt.
absolute counts (percentages). ‡Before first extubation attempt.

196.e4 Mastropietro et al