Tracheal intubation (TI) is a lifesaving procedure, yet it is known to carry a high risk of adverse events and worsened ICU outcomes (1 , 2). Hypoxemia is a commonly observed phenomenon during TI (3–8). Our group previously reported that the occurrence of hypoxemia or adverse events during TI is associated with multiple attempts, and significantly associated with duration of mechanical ventilation and length of ICU stay (6). The risk for hypoxemia during TI is higher in children due to lower functional residual capacity, predisposition to atelectasis as well as a higher oxygen utilization rate (9–12). The cause of hypoxemia during TI in children without heart disease or with noncyanotic heart disease is thought to be caused by a transient pulmonary venous desaturation due to increased intrapulmonary shunting (7 , 13). However, the hypoxemia during TI in children with cyanotic heart disease may be largely due to transient increase in intracardiac shunting (14–16). Therefore, the systemic effect of transient hypoxemia (i.e., oxygen desaturation) during TI may be different among children with cyanotic heart disease versus noncyanotic heart disease. Furthermore, it is unknown what degree of hypoxemia would occur with hemodynamically significant adverse TI associated event (TIAE) in children with cyanotic or noncyanotic heart disease such as cardiac arrest, hypotension, or dysrhythmia (17 , 18). It is possible that children with cyanotic heart disease may have a different threshold of systemic oxygen desaturation that places them at increased risk for hemodynamic TIAEs than patients with noncyanotic heart disease.
We sought to quantify the clinical impact of oxygen desaturation during TI in critically ill children with both cyanotic and noncyanotic heart disease. Our study hypothesizes that an association exists between oxygen desaturation during TI and the occurrence of hemodynamic TIAEs in both children with cyanotic heart disease and noncyanotic heart disease. We also hypothesize that the degree of oxygen desaturation from baseline associated with hemodynamic TIAEs differs between children with cyanotic heart disease and noncyanotic heart disease.
Study Population and Setting
We performed a retrospective analysis of prospectively collected observational data from the National Emergency Airway Registry for Children (NEAR4KIDS), an international multicenter quality improvement collaborative for airway management in critically ill children. We included all children (< 18 yr old) who underwent a primary TI in a pediatric or cardiac ICU between July 2012 and December 2016 at participating institutions, and who had a diagnosis of heart disease, either cyanotic or noncyanotic. This population included the data from 38 hospitals, the majority of which are located within the United States (Supplemental Appendix 1, Supplemental Digital Content 1, http://links.lww.com/PCC/A793). Each site leader was required to develop a site-specific compliance plan to ensure greater than 95% capture for all intubation encounters. These site-specific plans were approved by two NEAR4KIDS compliance officers prior to participation in the collaborative.
Institutional review board approval was obtained or waived as a quality improvement activity at each participating institution prior to the data collection for NEAR4KIDS. Data for the current study was extracted from the multicenter NEAR4KIDS database. Briefly, the dataset includes patient, provider, and practice characteristics associated with TI. The NEAR4KIDS data also includes TI processes and outcomes such as TIAEs. Data points used in this study included both cyanotic and noncyanotic heart disease as a primary ICU admission diagnosis, highest and lowest pulse oximetry value during the course of TI, and TI outcomes.
Definitions and Outcome Measures
The operational definitions developed and approved by the NEAR4KIDS Operational Definition Committee were followed throughout the data collection and verification at each site. Cyanotic heart disease was defined as the response given by NEAR4KIDS individual site personnel to the data collection form item with the following text, “Known cyanotic heart disease (R to L shunt)? (Circle ONE only) Yes/No.” These data point was verified by the site NEAR4KIDS leader or coordinator following the site-specific compliance plan but was not audited by the NEAR4KIDS data coordinating center. Our primary outcome was the occurrence of hemodynamic TIAEs which were defined as follows: cardiac arrest with or without return of spontaneous circulation (ROSC), hypotension requiring intervention, and dysrhythmia including bradycardia (heart rate < 60 beats/min) (6). Note that ongoing cardiac arrest and resuscitation before airway management was not considered as a hemodynamic TIAE (19).
Our primary exposure of interest was oxygen desaturation, which was defined as equal to or greater than a 15% drop from highest pulse oximetry value immediately before the intubation attempt. We further trichotomized the oxygen desaturation: 0–15%, 15% to less than 30%, and 30% or greater. The decision to trichotomization for the above cut off was made a priori based upon consensus by investigators. A sensitivity analysis was performed to explore other possible cut points using a spline analysis.
For summary statistics, categorical variables were described as number and percentages, and nonnormally distributed continuous variables as median and interquartile ranges. Baseline differences between children with cyanotic heart disease and noncyanotic heart disease were evaluated for categorical or dichotomous variables, and Wilcoxon rank-sum test was used for nonparametric variables. Next, we evaluated the occurrence of hemodynamic TIAEs in children with cyanotic and noncyanotic heart disease.
We assessed the predictive value of oxygen desaturation at two cut off levels (15% or greater and 30% or greater) for the occurrence of hemodynamic TIAEs in both children with cyanotic and noncyanotic heart disease. We calculated the diagnostic properties of each desaturation threshold (15% and 30%) and displayed positive predictive value, negative predictive value, sensitivity, and specificity. A multivariable logistic regression model was developed to evaluate the effect of desaturation at three levels (0% to <15%, 15% to < 30%, 30% or greater) on hemodynamic TIAEs for children with cyanotic and with noncyanotic heart disease separately. Covariates consistently associated with outcomes in the univariate analysis in both cohorts (p < 0.2) were included in these models. Cardiac medical versus surgical condition and provider level were chosen a priori as covariates for the models. The model fit was evaluated with the Hosmer-Lemeshow goodness of fit test. Once the model was developed, we estimated the marginal probability for hemodynamic TIAEs for each desaturation category. We repeated this multivariable analysis using different desaturation cut off levels (0–10%, 10–20%, 20–30%, and 30% or greater) as a sensitivity analysis. Given the arbitrary nature of these cutoffs, we ran the spline analysis to identify any data-driven cutoff points. All statistical analysis was performed using Stata 14.0 (Stata Corp, College Station, TX).
One-thousand nine-hundred ten children with cardiac disease who underwent TI were identified during the study period (cyanotic: n = 999; noncyanotic: n = 911). Patient demographics are shown in Table 1. Preprocedural oxygen saturations were lower in children with cyanotic heart disease compared with those with noncyanotic heart disease (median, 88%; 95% CI, 78–96% vs 100%; 95% CI, 95–100%; p < 0.001). Children who underwent TI with cyanotic heart disease were younger (p < 0.001) and more likely to have a cardiac-surgical diagnosis rather than a cardiac-medical diagnosis (p < 0.001). TIs in patients with cyanotic heart disease were significantly less likely to have hemodynamic failure (p = 0.009) or neurologic failure as a TI indication (p = 0.008). Patients with cyanotic heart disease were more likely to receive a vagolytic during the TI course (p < 0.001) and less likely to receive propofol (p = 0.001). In addition, children with cyanotic heart disease had higher Pediatric Index of Mortality 2 scores (p < 0.001). The vast majority (97%) of TIs in both groups had an attending level physician presence.
Oxygen Desaturation and Hemodynamic TIAEs
Oxygen desaturation was more commonly observed in children with cyanotic heart disease when compared with children with noncyanotic heart disease (23% vs 16% for desaturation between 15% to < 30%, and 23% vs 17% for desaturation equal to or greater than 30%; p < 0.001). There was no significant difference in the occurrence of hemodynamic TIAEs between patients with cyanotic and noncyanotic heart disease (7.5% vs 6.9%; p = 0.618) (Table 2).
Figure 1 demonstrates the association between each level of arterial oxygen desaturation and proportion of children who experienced hemodynamic TIAEs stratified by cyanotic and noncyanotic heart disease. We found the occurrence of hemodynamic TIAEs consistently increase with oxygen desaturation 30% or greater from baseline.
The predictive value for the occurrence of hemodynamic TIAEs in oxygen desaturation at the 15% and 30% cut off level were similar between children with cyanotic versus noncyanotic heart disease. For oxygen desaturation 15% to less than 30% the positive predictive value was 10% and 11% in children with cyanotic heart disease and noncyanotic heart disease, respectively and was 14% for oxygen desaturation greater than or equal to 30% in both groups.
Multivariable Logistic Regression Analysis
The multivariable logistic regression analyses for children with cyanotic and noncyanotic heart disease showed an increase in odds ratio for hemodynamic TIAEs increasing levels of arterial oxygen desaturation (Table 3 and Fig. 2). The odds ratio for hemodynamic TIAEs was significantly lower when the provider was a fellow performing the TI in children with both cyanotic and noncyanotic heart disease, both p value of less than 0.05. These models included patient age, medical versus surgical, hemodynamic instability, procedural indication, and provider level as covariates. The odds ratio at each cut off level was similar among children with cyanotic heart disease and with noncyanotic heart disease. The predicted probability of hemodynamic instability for both children with cyanotic and noncyanotic heart disease was 4.4% (95% CI, 3.2–5.6%) for oxygen desaturation 0 to less than 15%, 7.4% (95% CI, 4.7–10.2%) for desaturation 15 to less than 30%, and 14.3% (95% CI, 10.7–17.8%) for desaturation greater than or equal to 30%, as shown in Figure 2.
The spline analyses to identify data-driven cutoffs did not reveal any specific breakpoint with regards to the odds ratio for hemodynamic TIAEs (Supplemental Fig. 1, Supplemental Digital Content 2, http://links.lww.com/PCC/A794). We also repeated the main analysis using different oxygen desaturation cutoffs of less than 10%, 10–20%, 20–30%, above 30% with similar results (Supplemental Table 1, Supplemental Digital Content 3, http://links.lww.com/PCC/A795).
In our analyses with 1,910 children with cyanotic and noncyanotic heart disease who underwent TI, there was a significant increase in adverse hemodynamic TIAEs in both groups with a desaturation level greater than or equal to 30% compared with baseline saturation level. There was a rise in adverse event rates from the 0–15% desaturation group, to the 15% to less than 30% desaturation group and then to the 30% or greater desaturation group in both the cyanotic heart disease group and noncyanotic heart disease group. We did not identify a discrete threshold for oxygen desaturation for an increased risk for hemodynamic TIAEs. Although the categorization of cyanotic versus noncyanotic heart disease in our study was based on the site provider’s understanding, the preprocedure oxygen saturation level was distinctly different between the two groups, supporting the validity of the data. To our knowledge, the relationship between the degree of transient oxygen desaturation during TI and hemodynamic adverse events in children with acquired or congenital heart disease has not been previously reported.
In a previous study with the NEAR4KIDS database, Nishisaki et al (1) reported that an oxygen desaturation less than 80% in pediatric patients was associated with an increase in adverse TIAEs. However, this study population was not specific to cardiac patients and outcomes included all TIAEs which contained technical adverse events such as esophageal intubation or dental or lip trauma. Li et al (6) also reported that the occurrence of moderate desaturation defined as an arterial oxygen saturation less than 80% is common (19%) in all TIs in critically ill children, and that infant age, respiratory indication, and the use of a vagolytic agent and ketamine were associated with arterial oxygen desaturation. After adjusting for these confounding factors, moderate desaturation was significantly associated with hemodynamic TIAEs with odds ratio 1.83 (95% CI, 1.34–2.51; p < 0.001), and severe desaturation (defined as arterial oxygen saturation < 70%) was associated with hemodynamic TIAEs with an odds ratio of 2.16 (95% CI, 1.54–3.04; p < 0.001). Our study demonstrated a gradual increase in adverse hemodynamic TIAEs with the degree of oxygen desaturation and showed no absolute safe threshold. These study results suggest that prevention or reduction of oxygen desaturation during TI is important in order reduce these hemodynamic TIAEs. The NEAR4KIDS collaborative has instituted is an ongoing multicenter effort to reduce desaturation by implementing video laryngoscopy (to reduce multiple attempts) and use of an apneic oxygenation technique, a method to provide oxygen during laryngoscopy (2).
In this study, children with cyanotic heart disease were younger compared with those with noncyanotic heart disease. Infants are more prone to desaturation during apnea and have faster rates of onset of clinically significant hypoxemia (9). Interestingly, in the multivariable analysis, older age was associated with adverse hemodynamic events in the cardiac population as shown in Table 3. We speculate this result might be due to the complex physiology in older cardiac patients who underwent TIs in cardiac ICUs. For example, some of these patients might have chronic heart failure or Fontan physiology and thus are more susceptible to effects of positive pressure ventilation. Other patients may have coexisting pulmonary hypertension with high risk for pulmonary hypertension crisis during airway management. In addition, vagolytic agents were more frequently used in younger patients, which may or may not have contributed to this finding.
We found that while oxygen desaturation was more commonly observed in children who underwent TIs with cyanotic heart disease, the effect of hypoxemia on the occurrence of hemodynamic TIAEs was not significantly different between patients with cyanotic and noncyanotic heart disease. This finding requires further explanation. Given the preprocedural oxygen saturation level difference, the children with cyanotic heart disease had much more profound hypoxemia for the same level of oxygen saturation drop. Despite this, the occurrence of hemodynamic adverse events were similar. This may be due to a physiologic adaptation mechanism to acute desaturation after a chronic exposure or preconditioning to hypoxemia (20–24). Given those protective physiologic adaptations, it is possible that profound hypoxemia is better tolerated in children with cyanotic heart disease. The other explanatory factor is that the right to left shunt in cyanotic heart disease preserved systemic blood flow despite their more profound hypoxemia which was the key to sustain systemic circulation and avoid subsequent hemodynamic derangement. Additionally, more children in the noncyanotic group experienced hemodynamic TIAEs at lower changes in arterial oxygen saturations, and both groups had a linearly increasing event rate at more profound changes in arterial oxygen saturation, as shown in Figure 1.
Our current and previous study results show that cardiac arrest with or without ROSC was observed in 2.7% of TIs in children with heart disease (25). The hemodynamic effects of the intubation process on children with heart disease are complicated by physiologic factors such as shunting, myocardial dysfunction, outflow tract obstruction, conversion from negative to positive intrathoracic pressure, and the pharmacologic effects of induction and paralytic medications. Although no specific medications were significantly associated with the hemodynamic TIAEs in this study population, patient hemodynamic instability, procedural indication, and provider levels were associated with outcomes. On our multivariable analysis, we found a lower rate of TIAEs when the fellow was performing the TI compared with other provider levels. Since the vast majority of TIs were supervised by attending physicians, it is possible that high-quality patient management by an attending physician in addition to a proficient fellow level airway provider might be an actual contributing factor for lower hemodynamic TIAEs. TIs in children with cardiac disease should be approached cautiously by assessing these risk factors and addressing them when appropriate, including the risk for hypoxia. Matching provider skill set to the high-risk patient to optimize first attempt success is one way to minimize the risk of adverse hemodynamic events. An airway bundle checklist should be used to systematically address these risk factors and identify the safest approach (26).
Data from a registry is limited by reporting bias and inaccuracy of the data points used; however, in our case, this is minimized by the site-specific data compliance plans and universal operational definitions. The data coordinating center also performed periodic data quality checks. The categorization of cyanotic heart disease and noncyanotic heart disease was based on individual NEAR4KIDS site personnel’s understanding of the patient’s physiology at the time they underwent TI, as described in Methods. Formal auditing and harmonization with more precise anatomic and physiologic descriptors of Society of Thoracic Surgeons, the International Paediatric and Congenital Cardiac Code of the International Society for Nomenclature of Paediatric and Congenital Heart Disease, or Risk Adjustment for Congenital Heart Surgery scores (a scoring system to classify patients with congenital heart disease based on their predicted outcomes) were beyond the scope of this work (27 , 28). The investigators are aware that within a single disease name (e.g., Tetralogy of Fallot) that physiology can vary widely from pulmonary overcirculation (e.g., “Pink Tetralogy of Fallot”) to severe cyanosis and hypercyanotic spells warranting emergency intervention. Therefore, it is recognized that this categorization into cyanotic and noncyanotic is not calibrated to fully reflect nuances of individual patient physiology, nor exact descriptors of anatomic subtype, nor residual lesions. However, the broad dichotomization (into cyanotic vs noncyanotic) does reflect a commonly used construct for bedside teaching and clinical care, and our work includes detailed analysis above based upon baseline oxyhemoglobin saturation. Last, preprocedure oxygen saturation level likely reflected the preoxygenation. It would be helpful to know the patient’s oxygen saturation at baseline: that is, before preoxygenation was started. Unfortunately, this is not currently captured in the NEAR4KIDS database, but may warrant evaluation in future studies.
The level of hypoxemia and adverse events may not represent differences in only patient physiology but also TI process. One important aspect of that process not captured in our database is preoxygenation practice among children with cyanotic versus noncyanotic heart disease. For example, preoxygenation practice might be cautiously used in children with cyanotic heart disease with single ventricle physiology, whereas it is more likely used in children with noncyanotic heart disease.
Oxygen desaturation was commonly observed in children with both cyanotic and noncyanotic heart disease undergoing TI in pediatric and cardiac ICUs. The occurrence of hemodynamic TIAEs such as cardiac arrest, hypotension, and dysrhythmia were significantly associated with the occurrence of desaturation. The effect of oxygen desaturation during TI on adverse events was similar in both cyanotic and noncyanotic disease patients.
We thank all site team members in the National Emergency Airway Registry for Children collaboration, especially the site data coordinators and leaders for their diligent work in collecting the data. We thank Hayley Buffman, MPH, for administrative support, and Stephanie Tuttle, MBA, for her administrative leadership and support of this project throughout.
1. Nishisaki A, Turner DA, Brown CA III, et alNational Emergency Airway Registry for Children (NEAR4KIDS); Pediatric Acute Lung Injury and Sepsis Investigators (PALISI) Network: A National Emergency Airway Registry for Children: Landscape of tracheal intubation
in 15 PICUs. Crit Care Med 2013; 41:874–885
2. Parker MM, Nuthall G, Brown C III, et alPediatric Acute Lung Injury and Sepsis Investigators (PALISI) Network: Relationship between adverse tracheal intubation
associated events and PICU outcomes. Pediatr Crit Care Med 2017; 18:310–318
3. Lee JH, Turner DA, Kamat P, et alPediatric Acute Lung Injury and Sepsis Investigators (PALISI); National Emergency Airway Registry for Children (NEAR4KIDS): The number of tracheal intubation
attempts matters! A prospective multi-institutional pediatric observational study. BMC Pediatr 2016; 16:58
4. Wong DT, Yee AJ, Leong SM, et alThe effectiveness of apneic oxygenation during tracheal intubation
in various clinical settings: A narrative review. Can J Anaesth 2017; 64:416–427
5. Mosier JM, Joshi R, Hypes C, et alThe physiologically difficult airway. West J Emerg Med 2015; 16:1109–1117
6. Li S, Hsieh TC, Rehder KJ, et alFrequency of desaturation and association with hemodynamic adverse events during tracheal intubations in PICUs. Pediatr Crit Care Med 2018; 19:41–50
7. Mort TCComplications of emergency tracheal intubation
: Immediate airway-related consequences: Part II. J Intensive Care Med 2007; 22:208–215
8. Mort TCPreoxygenation in critically ill patients requiring emergency tracheal intubation
. Crit Care Med 2005; 33:2672–2675
9. Patel R, Lenczyk M, Hannallah RS, et alAge and the onset of desaturation in apnoeic children. Can J Anaesth 1994; 41:771–774
10. Rinderknecht AS, Mittiga MR, Meinzen-Derr J, et alFactors associated with oxyhemoglobin desaturation during rapid sequence intubation in a pediatric emergency department: Findings from multivariable analyses of video review data. Acad Emerg Med 2015; 22:431–440
11. Siddiqui SS, Janarthanan S, Harish MM, et alComplications of tracheal intubation
in critically ill pediatric cancer patients. Indian J Crit Care Med 2016; 20:409–411
12. Mort TCEmergency tracheal intubation
: Complications associated with repeated laryngoscopic attempts. Anesth Analg 2004; 99:607–613
13. Binks MJ, Holyoak RS, Melhuish TM, et alApnoeic oxygenation during intubation in the intensive care unit: A systematic review and meta-analysis. Heart Lung 2017; 46:452–457
14. Mojadidi MK, Ruiz JC, Chertoff J, et alPatent foramen ovale and hypoxemia. Cardiol Rev 2018 Mar 22
15. Mekontso Dessap A, Boissier F, Leon R, et alPrevalence and prognosis of shunting across patent foramen ovale during acute respiratory distress syndrome. Crit Care Med 2010; 38:1786–1792
16. Stark RJ, Shekerdemian LSEstimating intracardiac and extracardiac shunting in the setting of complex congenital heart disease
. Ann Pediatr Cardiol 2013; 6:145–151
17. Roppolo LP, Wigginton JGPreventing severe hypoxia during emergent intubation: Is nasopharyngeal oxygenation the answer? Crit Care 2010; 14:1005
18. Zeyneloglu P, Donmez A, Sener MSevoflurane induction in cyanotic and acyanotic children with congenital heart disease
. Adv Ther 2008; 25:1–8
19. Shiima Y, Berg RA, Bogner HR, et alNational Emergency Airway Registry for Children Investigators: Cardiac arrests associated with tracheal intubations in PICUs: A multicenter cohort study. Crit Care Med 2016; 44:1675–1682
20. Lim JM, Kingdom T, Saini B, et alCerebral oxygen delivery is reduced in newborns with congenital heart disease
. J Thorac Cardiovasc Surg 2016; 152:1095–1103
21. Wheaton WW, Chandel NSHypoxia. 2. Hypoxia regulates cellular metabolism. Am J Physiol Cell Physiol 2011; 300:C385–C393
22. Downing SE, Remensnyder JP, Mitchell JHCardiovascular responses to hypoxic stimulation of the carotid bodies. Circ Res 1962; 10:676–685
23. Cordina RL, Celermajer DSChronic cyanosis
and vascular function: Implications for patients with cyanotic congenital heart disease
. Cardiol Young 2010; 20:242–253
24. Rosove MH, Perloff JK, Hocking WG, et alChronic hypoxaemia and decompensated erythrocytosis in cyanotic congenital heart disease
. Lancet 1986; 2:313–315
25. Gradidge EA, Bakar A, Tellez D, et alNational Emergency Airway Registry for Children (NEAR4KIDS) and Pediatric Acute Lung Injury and Sepsis Investigators (PALISI): Effect of location on tracheal intubation
safety in cardiac disease-are cardiac ICUs safer? Pediatr Crit Care Med 2018; 19:218–227
26. Li S, Rehder KJ, Giuliano JS Jr, et alNational Emergency Airway Registry for Children (NEAR4KIDS) Investigators; Pediatric Acute Lung Injury and Sepsis Investigator PALISI Network Investigators: Development of a quality improvement bundle to reduce tracheal intubation
-associated events in pediatric ICUs. Am J Med Qual 2016; 31:47–55
27. Jenkins KJRisk adjustment for congenital heart surgery: The RACHS-1 method. Semin Thorac Cardiovasc Surg Pediatr Card Surg Annu 2004; 7:180–184
28. Jacobs JP, O’Brien SM, Pasquali SK, et alThe society of thoracic surgeons congenital heart surgery database mortality risk model: Part 2-clinical application. Ann Thorac Surg 2015; 100:1063–1068; discussion 1068–1070
adverse event; congenital heart disease; cyanosis; oxygen desaturation; tracheal intubation
Supplemental Digital Content
©2019The Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies