Extubation failure is associated with significant morbidity and mortality in ICU patients (1 , 2). Noninvasive ventilation (NIV), including high-flow nasal cannula (HFNC) and positive airway pressure (PAP) support, is often employed after extubation of high-risk patients to mitigate the risk of extubation failure (3). NIV has become common in the postoperative management of pediatric cardiac surgical patients, despite a paucity of data demonstrating clinical benefits (4 , 5). Indiscriminate use of NIV and/or slow weaning from NIV may prolong the ICU exposure period, increasing the risk for healthcare-associated complications. Furthermore, NIV may be associated with an increased risk of aspiration and feeding intolerance (6 , 7), often leading clinicians to be more judicious with enteral feed advancement. The result may be inadequate nutrition delivery and delayed oral feeding and delayed postoperative convalescence (8–10).
Studies have demonstrated that HFNC is capable of producing similar airway pressures as continuous PAP (CPAP) and bilevel PAP (BiPAP) (11 , 12), which has led it to be classified as NIV in previous ICU publications (5 , 13). In our center, use of noninvasive PAP is associated with higher daily cost and resource utilization as it requires a mechanical ventilator and frequent respiratory therapist involvement, whereas HFNC is titrated by the bedside nurse. Lack of definitive clinical benefit of PAP compared with HFNC in the postoperative management of pediatric cardiac ICU (CICU) patients has generated clinical equipoise regarding the primary choice for postextubation respiratory support selection in high-risk CICU patients.
In this study, we sought to describe the impact of postsurgical respiratory support with PAP versus HFNC in infants with congenital heart disease. Our primary aim was to explore the impact of initial respiratory modality on extubation failure rate. Secondarily, we also aimed to evaluate the association of postextubation respiratory support mode with postsurgical resource utilization, including total duration of respiratory support (i.e., time to achieving low-flow nasal cannula [LFNC] and room air). We hypothesized that extubation failure rate would be similar between the two modalities and that HFNC would be associated with decreased CICU resource utilization.
MATERIALS AND METHODS
We performed a retrospective cohort study of patients less than 6 months old admitted to the Children’s of Alabama CICU between July 1, 2012, and June 30, 2015, after congenital heart surgery requiring cardiopulmonary bypass (CPB) through an open sternotomy. Exclusion criteria included postoperative extracorporeal cardiopulmonary support, tracheostomy tube cannulation prior to surgery, orders to limit life-sustaining therapy, transfer out of the CICU prior to extubation, or death prior to extubation. All data were collected from patient electronic medical records and our CICU clinical database. The University of Alabama at Birmingham institutional review board approved this study and waived the need for informed consent.
Respiratory Modalities and Equipment
PAP included CPAP with or without pressure support or BiPAP through a Neotech (Valencia, CA) RAM cannula or nasal mask system (Fisher & Paykel, Auckland, New Zealand) delivered by a Servo-i ventilator (Maquet, Rastatt, Germany). HFNC oxygen support included humidified oxygen delivered at 3 liters per minute (LPM) or greater through an Optiflow Junior (Fisher & Paykel, Auckland, New Zealand) HFNC set-up using an MR850 (Fisher & Paykel, Auckland, New Zealand) heated humidifier. LFNC oxygen support was defined as less than or equal to 2 LPM of humidified oxygen delivered by an AirLife (CareFusion, Yorba Linda, CA) nasal cannula. Our heart center exclusively used cuffed endotracheal tubes and an invasive mechanical ventilation (MV) weaning protocol during the study period.
Postextubation respiratory support selection was made at the discretion of the attending CICU physician. Generally, our institutional HFNC practice started flows of 3–6 LPM in neonates and 3–8 LPM in infants. HFNC flow was increased to a maximum of 8 LPM in neonates and 12 LPM in infants as clinically indicated before declaring this modality a treatment failure. PAP was generally initiated with BiPAP using an inspired PAP (iPAP) of 20 cm H2O, expired PAP (ePAP) of 10 cm H2O, and supported rate of 20–30. PAP was increased to maximum settings of iPAP 25 cm H2O, ePAP 15 cm H2O, and supported rate of 30 as clinically indicated before declaring this modality a treatment failure. FIO2 was titrated for both HFNC and PAP, as appropriate, at the discretion of the treating clinician. Treatment failure was determined clinically by observing signs of increased work of breathing (e.g., tachypnea, grunting, accessory muscle use, irritability), declining renal near-infrared spectroscopy reading, increased arteriovenous oxygen and/or carbon dioxide gradient, and/or increased serum lactate measurement. For patients failing HFNC, a trial of PAP was considered before tracheal reintubation. For patients failing PAP, endotracheal reintubation was performed.
Weaning of postsurgical HFNC and PAP was not protocolized. In general, HFNC was weaned to LFNC at the discretion of the bedside nurse based upon hourly patient assessments and daily physiologic goals. The decision to wean PAP was driven by respiratory therapist assessment of the patient’s respiratory status every 4–6 hours and physician assessment of the patient’s clinical trajectory. Generally, PAP was weaned from BiPAP to CPAP, followed by a transition to HFNC; however, some patients were transitioned to from CPAP to LFNC as clinically indicated.
Neonates were defined as patients with a chronological age less than 30 days on the day of cardiac surgery. Infants were defined as patients with a chronological age of 30–364 days on the day of cardiac surgery. First planned postoperative extubation in the CICU was considered the index extubation. Initial postextubation respiratory support was defined as the first mode of support documented after index extubation. Airway anomalies included congenital or acquired upper or lower airway abnormalities (e.g., cleft lip and/or palate, laryngeal web, laryngomalacia, postintubation subglottic stenosis, tracheo/bronchomalacia, congenital lobar emphysema). Diaphragm paresis was determined by ultrasonography or fluoroscopy. Vocal cord paresis was defined as weakness or immobility on fiberoptic endoscopic evaluation of swallow. Nitric oxide use within 12 hours before or after index extubation was a surrogate for pulmonary hypertension. Dexamethasone use within 12 hours before or after the index extubation was a surrogate for periextubation upper airway obstruction. Periextubation sedation use was defined as dexmedetomidine infusion or “scheduled” lorazepam, phenobarbital, or methadone administered within 12 hours before or after index extubation. Rescue PAP was defined as transition of a patient to PAP after initial extubation to HFNC. Extubation failure was defined as need for tracheal reintubation within 48 hours of index extubation.
Enteric tube feeding management during PAP support was at the discretion of the CICU attending. Our CICU feeding protocols dictate delay of oral feeding until infants were on LFNC support. First oral feed attempt was defined as the infant’s first recorded oral feeding of any volume administered by a nurse, speech therapist, or occupational therapist. For infants at high risk for aspiration or with poor oral skills, protocolized modified barium swallow (MBS) evaluation preceded first oral attempt and was thus used as a surrogate for first oral feed attempt as indicated (for aspiration risk factors used by our CICU to determine need for swallow evaluation, see Supplemental Appendix, Supplemental Digital Content 1, http://links.lww.com/PCC/A821). Goal bolus feeding was defined as the first charted bolus feed volume of 15 cc/kg or greater by mouth or gastrostomy tube (GT) (14). Our institutional GT insertion practice was guided by a team approach: after multiple failed attempts at oral feeding with the bedside nurse and speech therapist and at least two failed MBS evaluations, the treating clinician approached the patient’s family for GT insertion by a pediatric general surgeon to optimize nutrition delivery.
Preoperative data collected included chronological age on the day of surgery (d); CICU admission weight (kg) and weight-for-age z score (15); preoperative intubation; and preoperative single ventricle physiology. Operative and postoperative data included Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery (STAT) mortality category (16); CPB duration (min); postoperative MV duration (hr) prior to index extubation; Vasoactive-Inotropic Score (17) at the time of extubation; and periextubation administration of inhaled nitric oxide, sedation, or dexamethasone. Airway anomalies, diaphragm paresis, vocal cord paresis known before or after cardiac surgery were recorded. Postextubation respiratory support data included duration of initial PAP support; initial HFNC oxygen flow rate and FIO2 applied immediately after index extubation; number of subjects escalated from HFNC to PAP (rescue PAP); and duration of rescue PAP.
Patients initially extubated to PAP after index extubation were compared with patients extubated to HFNC. Primary outcomes included extubation failure rate within 48 hours of index extubation; time from index extubation to extubation failure and in-hospital mortality were also reported. Surrogates for resource utilization that were measured included the following: total durations of postoperative MV, PAP, HFNC, and LFNC; time from index extubation to LFNC and room air; time from CICU postoperative admission to first oral feed attempt and goal bolus feeding; frequency of GT insertions during hospitalization after index extubation; and postsurgical hospital length of stay (HLOS). Postsurgical HLOS was used instead of CICU length of stay (LOS) given the colinearity of CICU LOS with postsurgical HLOS in addition to our practice of discharging the more complex congenital heart surgery patients to home directly from the CICU.
Covariates and outcomes for the two treatment groups before propensity matching were compared using Pearson chi-square, Fisher exact, or Mann-Whitney U tests, as appropriate, for categorical or continuous variables. Inherent in this study was the clear selection bias toward increased postextubation respiratory support in the patients perceived to be at highest risk for extubation failure. To reduce treatment selection bias and derive a reasonably balanced population for comparison, we performed a propensity-matched analysis after assembling a balanced cohort through propensity score estimation. Propensity scores were estimated by incorporating covariates likely to influence postextubation treatment selection or extubation failure risk into a multivariable logistic regression model using treatment selection as the outcome variable. We then performed one-to-one matching without replacement using the estimated propensity scores in each treatment assignment by employing nearest neighbor-matching procedures with a caliper width of 0.15 (18–20). We assessed postmatch balance of covariates by absolute standardized differences (20–23) with absolute standardized differences less than 0.20 indicating small or negligible imbalance between treatment groups (22). McNemar and Wilcoxon signed rank tests were used, as appropriate, for matched comparison of categorical and continuous covariates and outcome measures, respectively. The Mann-Whitney U test was used for comparison of unmatched categorical or continuous outcomes. Generalized estimating equation methods were employed to accommodate the correlation of observations due to the matched-pair nature of the data. Propensity matching and statistical analyses were completed using Statistical Package for Social Sciences, Version 23 (SPSS, Chicago, IL); R Version 3.2.4 (R Foundation for Statistical Computing, Vienna, Austria; available through https://cran.r-project.org/mirrors.html); and SAS, Version 9.4 (SAS Institute, Cary, NC). All statistical tests were two-tailed, and p values of less than 0.05 were considered significant.
Of 514 CICU encounters for congenital heart surgery during the study period, 260 (245 patients) met inclusion criteria for analysis. Table 1 displays patient characteristic and management variables for the HFNC and PAP groups. Patients extubated to PAP were younger and had a higher frequency of preoperative invasive MV, higher STAT category operations, and longer CPB runs. Postoperatively, PAP subjects were supported with invasive MV for a longer duration, had more vocal cord paresis, and were treated more frequently with periextubation scheduled sedation and dexamethasone.
One-hundred–two patients (39%) were initially extubated to PAP. Neonates were more frequently extubated to PAP than infants (neonates vs infants; 62/110 [56%] vs 40/150 [27%]; p < 0.001). Median PAP duration was 45 hours (interquartile range [IQR], 25–88 hr). One-hundred fifty-eight patients (61%) were extubated to HFNC with a median initial flow of 5 LPM (IQR, 4–6 LPM) and FIO2 of 1.0 (IQR, 0.41). Thirty-five HFNC patients (22% of HFNC group) received rescue PAP at a median 10 hours (IQR, 4–24 hr) after extubation. Rescue PAP was used for a median 45 hours (IQR, 22–74 hr). Patients who received rescue PAP were younger; underwent longer, more complex surgeries; were supported with invasive MV a median 8 hours longer; and had more airway anomalies and diaphragm paresis than patients who were supported with HFNC alone (Supplemental Table 1, Supplemental Digital Content 2, http://links.lww.com/PCC/A822).
Unmatched outcomes are presented in Supplemental Table 2 (Supplemental Digital Content 3, http://links.lww.com/PCC/A823). In summary, patients extubated to PAP required 3 and 5 days longer to wean to LFNC and room air, respectively (p < 0.001 for all comparisons), compared with HFNC. PAP use was also associated with increased time to achieve nutritional goals and longer postsurgical HLOS (p < 0.001 for all comparisons). Extubation failure rate was 12% (31/260) and did not significantly differ between respiratory support groups; indications were respiratory failure (n = 21), shock (n = 6), cardiac arrest (n = 3), and altered mental status (n = 1). Patients who failed extubation were younger, underwent more complex surgeries and experienced more diaphragm paresis than patients who successfully extubated (Supplemental Table 3, Supplemental Digital Content 4, http://links.lww.com/PCC/A824).
Of the two-hundred fifty-eight patients considered for propensity score matching (two were excluded for missing data), we were able to match 49 pairs. There were no significant residual differences in patient demographics or treatment factors between the HFNC and PAP groups after matching (Table 1). Standardized differences between matched covariates were well balanced (Fig. 1).
Extubation failure rate did not significantly differ between the HFNC and PAP groups (HFNC vs PAP; 5/49 [10%] vs 8/49 [16%]; p = 0.549) (Table 2), nor did the time to extubation (HFNC vs PAP; 26 hr [IQR, 3–29 hr] vs 3 hr [IQR, 1–15 hr]; p = 0.284). Fifteen HFNC patients (31%) received rescue PAP a median 10 hours (IQR, 6–25 hr) after extubation for a median duration of 50 hours (IQR, 28–115 hr). Three patients (20%) who received rescue PAP failed extubation at a median 28 hours (IQR, 26–29 hr).
The PAP group took approximately 2 days longer to wean to both LFNC and room air (p = 0.006 and 0.013, respectively) (Table 2) that was related to longer durations of PAP and LFNC in this group. First oral feed attempt or MBS evaluation for oral feeding readiness was delayed by almost 4 days in the PAP group compared with HFNC (p = 0.034). Extubation to PAP was also associated with a delay of approximately 3 days to reach goal bolus feeding compared with HFNC (p = 0.020). There were more GT insertions in the PAP group (PAP vs HFNC; 18/42 [43%] vs 8/46 [17%]; p = 0.009). PAP was associated with an increase in median postsurgical HLOS by 8 days (p = 0.015).
NIV is commonly used after congenital heart surgery in an attempt to mitigate extubation failure risk and improve other respiratory-related clinical outcomes (4 , 5); however, evidence to guide NIV mode selection and support the use of PAP over HFNC is lacking. In this study, we retrospectively evaluated our institutional practice of initial respiratory support selection for neonates and infants up to 6 months old extubated after congenital heart surgery. In a matched cohort, we found extubation to PAP compared with HFNC oxygen support was associated with increased postsurgical resource utilization, evidenced by longer time to wean to LFNC and room air, time to goal enteral nutrition variables, and HLOS, without a difference in extubation failure rate. These findings suggest that, when applied prophylactically and broadly after initial extubation in neonates and infants, PAP use appears to increase resource utilization and exposure to ICU care without a demonstrable outcome benefit when compared with HFNC.
Extubation failure after congenital heart surgery occurs in 6–12% of planned extubations and is associated with prolonged ICU LOS and in-hospital mortality (24–26). Our overall extubation failure rate was 12% with extubation failure occurring more frequently in younger patients (median age, 18 d) who underwent more complex cardiac surgeries and experienced more diaphragm paresis. Like many centers, PAP was applied to patients in our CICU at the discretion of the attending for perceived risk of extubation failure. In our matched cohort, PAP use was not associated with reduced extubation failure compared with HFNC oxygen support. Three points must be considered when interpreting these results. First, both HFNC and PAP may deliver similar levels of positive pressure. HFNC has been shown to produce variable levels of PAP in neonates (27), and PAP delivered nasally may be lost as a result of inadequate nasal antral seal and mouth-opening. Second, 15 patients (31%) in the HFNC group received rescue PAP shortly (median, 10 hr) after extubation, of whom only three patients required reintubation; thus, it is possible that some of the patients in the HFNC group realized treatment benefit from PAP therapy with respect to extubation failure rate. This may also help explain the absolute difference in time to extubation failure between the matched groups as three of the five patients who failed extubation in the HFNC group were placed on rescue PAP before eventual reintubation, thus giving the appearance that patients supported with HFNC fail extubation later than those supported by PAP. Third, there were patients in both respiratory support groups who may have not needed any form of NIV support to successfully transition from invasive ventilation, thereby potentially diluting the treatment effect seen from either modality. However, without objective data (e.g., lung compliance information, cardiac function analysis, diaphragm paresis screening) to more carefully discern the patients that may have successfully extubated without HFNC or PAP, this point remains a conjecture.
For clinicians who routinely provide perioperative care for pediatric patients with congenital heart disease, the association of postextubation PAP use with increased resource utilization when compared with HFNC may not be entirely surprising. There are meaningful differences in the use of PAP and HFNC that could account for the outcomes we found. At our institution, application of PAP requires a ventilator or CPAP device and associated equipment, whereas HFNC uses much of the same equipment as LFNC. It is likely that clinician preference for PAP, due to the perception that it provides a higher level of respiratory support than HFNC in this high-risk population, and the lack of a PAP weaning protocol at our institution significantly influenced the duration of total respiratory support in the PAP cohort. Additionally, common CICU clinician management is to empirically wean PAP support and then transition to HFNC to complete weaning to LFNC. It is possible that many of these children could have been weaned directly from low settings on PAP to LFNC.
Postoperative PAP may benefit pediatric CICU patients via decreased left ventricular afterload, maintenance of functional residual capacity with reduction of pulmonary vascular resistance and right ventricular afterload, and decreased oxygen consumption by respiratory muscles (28). Although initial selection of PAP after extubation may be entirely appropriate, patients with congenital heart disease who are treated with PAP are often perceived to be particularly fragile, which may bias physician behavior toward more cautious weaning. At our institution, PAP weaning is entirely dependent upon clinician directive. Objective assessment for PAP weaning-readiness by our clinicians may occur infrequently, leading to unnecessarily prolonged PAP support. In contrast, patients supported with HFNC may be viewed as less fragile and thus are allowed to undergo faster weaning of respiratory support by the bedside nurse based upon hourly assessments of physiologic and goal-direct variables agreed upon during morning rounds. Thus, the respiratory support duration benefit realized by the HFNC group may largely be a product of nurse-driven HFNC weaning. Although we cannot conclude which postoperative cohorts would most benefit from PAP based on our study, we believe that initiatives guiding more discriminant PAP use and nurse- or respiratory therapist-driven PAP weaning protocols could expedite postoperative convalescence (29 , 30) and potentially reduce resource utilization.
Adequate nutrition is critical to postsurgical recovery (31). In our matched cohort, we found that patients extubated to HFNC attempted oral feeds a median 4 days sooner and achieved goal bolus feeding a median 3 days sooner than PAP. Furthermore, despite the use of a detailed postoperative feeding protocol at our institution, we saw consistent evidence of enteral feeding interruption in PAP patients regardless of the presence of a transpyloric feeding tube (data not shown). Although we cannot conclude that the relationship between PAP use and feeding delays is causal, it is dogma that the use of PAP may preclude more aggressive oral feeding. In our experience, clinicians often stop or decrease volume of enteral feeds in infants supported with PAP and will not consider oral feeding until the patient is supported on LFNC supplemental oxygen given the perceived risk of extubation failure or aspiration. We conjecture that a potential unintended consequence of withholding oral feeding during PAP is exacerbation of feeding aversion or oral dyscoordination frequently seen after neonatal cardiac surgery (9 , 10). However, the higher frequency of postoperative vocal cord paresis in the matched PAP group compared with HFNC, although not statistically significant, would suggest that there may be other variables that could be co-contributors to the delay in oral feeding readiness. Nonetheless, the combination of inconsistent delivery of enteral feeds and delayed oral feeding in the PAP group potentially contributed to the higher rate of GT insertions and longer postsurgical HLOS we saw. Ultimately, initiatives aimed at defining best practice for PAP nutrition and oral feeding readiness are needed to inform standardized feeding practices during PAP that could enable attainment of adequate nutrition goals and oral feeding more consistently.
Despite our findings, we surmise extubation to PAP is appropriate for select postoperative cardiac patients. We base this opinion on the observation that extubation failure was not significantly different between unmatched respiratory support groups despite a clear severity of illness difference between the groups, suggesting that there were patients in the PAP group who benefited from the support they received. Although neonatal studies suggest the noninferiority of HFNC in ameliorating extubation failure risk compared with PAP (32 , 33), we speculate that these studies may not directly apply to an infant after heart surgery. PAP theoretically reduces left ventricular afterload, improves functional residual capacity, and decreases oxygen demand more so than HFNC, which should particularly benefit the delicate cardiopulmonary physiology of the highest risk neonates and infants after heart surgery (28). As such, the results of our study should be interpreted with caution. As 53% of our highest surgical risk patients (i.e., STAT 4–5) were excluded from the propensity-matched analysis, our results may not directly apply to the most fragile postoperative pediatric cardiac patients. However, at present no objective data exist (e.g., work of breathing assessments, esophageal manometry, or oxygen consumption data) to substantiate the theoretical benefit of PAP over HFNC in neonates and infants after the most complex heart surgeries. Further study is ultimately needed to inform optimal patient selection for postextubation PAP or HFNC use and improve weaning after congenital heart surgery (34).
Our study has several important limitations. First, the study was retrospective and observational, intended only to evaluate the impact of CICU NIV modality selection on outcomes as it related to the initial management choice after postoperative extubation in patients who underwent cardiac surgeries requiring CPB. We cannot conclude anything further regarding the effectiveness of HFNC versus PAP at any other time point in the CICU stay or in any other critically ill pediatric population. Although assumed to be largely prophylactic, given the retrospective nature of the study, we could not control for the indication for respiratory support selection. Propensity analysis identified only 49 matched pairs, increasing the potential for type II error regarding extubation failure outcomes. Residual bias may have persisted after propensity matching as there were likely unaccounted variables that could have influenced the choice of respiratory support. Even after matching, there was an 11-day difference in median patient age at the day of surgery and a 15-hour difference in median postoperative MV duration between respiratory support groups, suggesting that there may exist residual unaccounted confounding variables (e.g., periextubation fluid balance, pulmonary compliance, rhythm disturbance, and cardiac dysfunction). Furthermore, the absolute difference in the median time to extubation failure between the matched groups, although not statistically significant, suggests the presence of unmeasured variables in the PAP group compared with HFNC. External validity of our study results is limited given the known variability of PAP utilization among centers (5) and our institutional case-mix, postextubation respiratory support selection and weaning practice, and feeding protocol. Finally, we also did not measure specific nutritional outcomes outside of enteral nutrition administration goals, which limits our conclusions regarding how delay in nutrition delivery affected postsurgical management.
In a propensity-matched study of postoperative neonates and infants less than 6 months old with congenital heart disease, primary extubation to PAP support, when compared with HFNC, was associated with greater postsurgical respiratory resource utilization, delay in nutrition delivery, and longer HLOS without a significant benefit in extubation failure rate. Multicenter prospective initiatives directed at establishing best clinical practice for postoperative NIV patient selection, timing, modality, weaning, and concurrent nutrition delivery are needed.
1. Kurachek SC, Newth CJ, Quasney MW, et al. Extubation failure in pediatric intensive care: A multiple-center study of risk factors and outcomes. Crit Care Med 2003; 31:2657–2664
2. Harkel AD, van der Vorst MM, Hazekamp MG, et al. High mortality rate after extubation failure after pediatric cardiac surgery. Pediatr Cardiol 2005; 26:756–761
3. Kovacikova L, Skrak P, Dobos D, et al. Noninvasive positive pressure ventilation in critically ill children with cardiac disease. Pediatr Cardiol 2014; 35:676–683
4. Fernández Lafever S, Toledo B, Leiva M, et al. Non-invasive mechanical ventilation after heart surgery in children. BMC Pulm Med 2016; 16:167
5. Romans RA, Schwartz SM, Costello JM, et al. Epidemiology of noninvasive ventilation
in pediatric cardiac ICUs. Pediatr Crit Care Med 2017; 18:949–957
6. Hanin M, Nuthakki S, Malkar MB, et al. Safety and efficacy of oral feeding in infants with BPD on nasal CPAP. Dysphagia 2015; 30:121–127
7. Ferrara L, Bidiwala A, Sher I, et al. Effect of nasal continuous positive airway pressure on the pharyngeal swallow in neonates. J Perinatol 2017; 37:398–403
8. Katona P, Katona-Apte J. The interaction between nutrition and infection. Clin Infect Dis 2008; 46:1582–1588
9. Kogon BE, Ramaswamy V, Todd K, et al. Feeding difficulty in newborns following congenital heart surgery. Congenit Heart Dis 2007; 2:332–337
10. Schwalbe-Terilli CR, Hartman DH, Nagle ML, et al. Enteral feeding and caloric intake in neonates after cardiac surgery. Am J Crit Care 2009; 18:52–57
11. ten Brink F, Duke T, Evans J. High-flow nasal prong oxygen therapy or nasopharyngeal continuous positive airway pressure for children with moderate-to-severe respiratory distress?*. Pediatr Crit Care Med 2013; 14:e326–e331
12. Metge P, Grimaldi C, Hassid S, et al. Comparison of a high-flow humidified nasal cannula to nasal continuous positive airway pressure in children with acute bronchiolitis: Experience in a pediatric intensive care unit. Eur J Pediatr 2014; 173:953–958
13. DeMauro SB, Millar D, Kirpalani H. Noninvasive respiratory support for neonates. Curr Opin Pediatr 2014; 26:157–162
14. Moellinger AB, Torsch S, Borasino S, et al. Postoperative feeding protocol improves outcomes after arterial switch operation. 2014;5:In: Select abstracts from Cardiology 2014: The 17th annual update on pediatric and congenital cardiovascular disease, February 19–23, 2014. Orlando, FL, 178World J Pediatr Congenit Heart Surg.
16. Jacobs ML, O’Brien SM, Jacobs JP, et al. An empirically based tool for analyzing morbidity associated with operations for congenital heart disease
. J Thorac Cardiovasc Surg 2013; 145:1046–1057
17. Gaies MG, Gurney JG, Yen AH, et al. Vasoactive-inotropic score as a predictor of morbidity and mortality in infants after cardiopulmonary bypass. Pediatr Crit Care Med 2010; 11:234–238
18. Rosenbaum PR, Rubin DB. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat 1985; 39:33–38
19. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011; 46:399–424
21. Flury BK, Riedwyl H. Standard distance in univariate and multivariate analysis. Am Stat 1986; 40:249–251
22. Normand ST, Landrum MB, Guadagnoli E, et al. Validating recommendations for coronary angiography following acute myocardial infarction in the elderly: A matched analysis using propensity scores. J Clin Epidemiol 2001; 54:387–398
23. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 1988Second Edition. Hillsdale, NJ, Lawrence Erlbaum Associates.
24. Gaies M, Tabbutt S, Schwartz SM, et al. Clinical epidemiology of extubation failure in the pediatric cardiac ICU: A report from the pediatric cardiac critical care consortium. Pediatr Crit Care Med 2015; 16:837–845
25. Benneyworth BD, Mastropietro CW, Graham EM, et al. Variation in extubation failure rates after neonatal congenital heart surgery across Pediatric Cardiac Critical Care Consortium hospitals. J Thorac Cardiovasc Surg 2017; 153:1519–1526
26. Mastropietro CW, Cashen K, Grimaldi LM, et al. Extubation failure after neonatal cardiac surgery: A multicenter analysis. J Pediatr 2017; 182:190–196
27. Spence KL, Murphy D, Kilian C, et al. High-flow nasal cannula as a device to provide continuous positive airway pressure in infants. J Perinatol 2007; 27:772–775
28. Cheifetz IM, Martin LD, Meliones JN, et al. Nichols DG, Ungerleider RM, Spevak PJ, et al. Respiratory support for the child with critical heart disease. In: Critical Heart Disease in Infants and Children. 2006, pp Second Edition. Philadelphia, PA, Mosby Elselvier, 307–332
29. Tonnelier JM, Prat G, Le Gal G, et al. Impact of a nurses’ protocol-directed weaning procedure on outcomes in patients undergoing mechanical ventilation for longer than 48 hours: A prospective cohort study with a matched historical control group. Crit Care 2005; 9:R83–R89
30. Kirakli C, Ediboglu O, Naz I, et al. Effectiveness and safety of a protocolized mechanical ventilation and weaning strategy of COPD patients by respiratory therapists. J Thorac Dis 2014; 6:1180–1186
31. Leong AY, Field CJ, Larsen BM. Nutrition support of the postoperative cardiac surgery child. Nutr Clin Pract 2013; 28:572–579
32. Manley BJ, Owen LS, Doyle LW, et al. High-flow nasal cannulae in very preterm infants after extubation. N Engl J Med 2013; 369:1425–1433
33. Yoder BA, Stoddard RA, Li M, et al. Heated, humidified high-flow nasal cannula versus nasal CPAP for respiratory support in neonates. Pediatrics 2013; 131:e1482–e1490
34. Inata Y, Takeuchi M. Complex effects of high-flow nasal cannula therapy on hemodynamics in the pediatric patient after cardiac surgery. J Intensive Care 2017; 5:30