Patient Characteristics Associated with Complications During Neonatal Intubations

Abstract

Introduction:

Complications of neonatal intubation are known to be increased with emergent intubations, increased number of attempts, unstable hemodynamics, or ventilation failure; and decreased with use of paralytic medication and videolaryngoscopy. Patient characteristics associated with complications are not well understood.

Design/Methods:

A retrospective cohort study was performed of neonates who underwent intubation between January 2017 and June 2019. Patient characteristics of infants with and without complications were compared. Complications included common adverse events and abnormal vital sign changes occurring during intubation.

Results:

A total of 467 intubation encounters in 352 infants were included with median gestational age (GA) at birth of 29 weeks, postmenstrual age (PMA) 33 weeks at intubation, and median weight 1795 grams. 41.5% of infants had complications and 58.5% of infants did not. Infants with complications compared to infants without had a median FiO₂ of 0.50 vs. 0.45 (p = 0.183), median GA at birth of 29 vs. 31 weeks (p<0.001), median PMA of 32 weeks vs. 33.0 weeks (p = 0.352), median weight of 1540g vs. 1970g (p = 0.091), and median chronological age of 3 days vs. 1 day (p = 0.001). Generalized Estimating Equations controlling for administration of paralytic indicated decreased complications in infants ≤21.5 days in chronological age (OR 0.45 95% CI 0.30-0.69) and increased complications in infants ≤1565g (OR 1.52 95% CI 1.04-2.23).

Conclusion:

Patient characteristics associated with an increased rate of complications included chronological age and weight. Further study is needed to reduce complications.

Introduction:

Neonatal endotracheal intubation is a lifesaving procedure for many neonates. Neonates in the intensive care unit frequently require intubation for respiratory failure, surfactant administration, elective procedures, unplanned extubations, or apnea. Additionally, intubation is a distressing and painful procedure with the potential for airway injury, physiologic derangement, and destabilization of the patient^1–3.

Complications related to intubation in the neonatal population have only recently been described and have a relatively high incidence. Previous studies have examined common adverse events and vital sign changes separately. Foglia et al described a large multi-center cohort of neonatal intubations showing a 22% rate of tracheal intubation related adverse events, and a 51% rate of desaturation more than 20% below baseline¹. In that study factors associated with increased risk of tracheal intubation related adverse events included provider training level and use of paralytic medications¹. Hatch et al described 178 patients undergoing intubation and found 39% rate of adverse events and a 44% rate of hypoxemia (oxygen saturation <60%) . That study showed increased adverse events with increasing number of attempts and an emergent setting compared to an elective or urgent setting². Studies in other populations have also shown that adverse events decrease with an increase in provider training level⁴, and are worse in emergent situations and after normal business hours⁵. Previous research has identified modifiable risk factors to decrease risk, including experience level of provider¹, videolaryngoscopy⁶, and use of premedication^{1; 6; 7}. This study adds to this body of literature specifically examining patient characteristics which may help identify those who are at high risk for complications during intubation. This study includes vital sign changes such as bradycardia and desaturation within the definition of complications; in contrast to previous studies. This information could help guide patient care to target interventions towards that population.

The study aim was to describe the incidence and type of complications during neonatal intubation and evaluate patient characteristics associated with complications; including fraction of inspired oxygen (FiO₂), carbon dioxide level (PCO₂), gestational age (GA) at birth, postmenstrual age (PMA), and weight. The authors hypothesized that a higher FiO₂, higher PCO₂, lower GA at birth, lower PMA, and lower weight would be associated with an increased odds of complications.

Materials/Methods:

This study is a retrospective cohort study examining the patient characteristics associated with complications during neonatal intubation. This study was approved by the University Hospitals Institutional Review Board. At this institution, data on complications during neonatal intubations have been monitored prospectively since 2017 as part of ongoing quality improvement efforts⁸. A quality data sheet is filled out on intubations occurring in the neonatal intensive care unit (NICU) by the bedside nurse at the time of intubation.

The setting is an 80-bed academic level IV NICU. Providers performing intubation include neonatal-perinatal medicine fellows, neonatal nurse practitioners, pediatric residents, neonatal-perinatal medicine attending physicians, respiratory therapists, and occasionally subspecialists such as otolaryngologists or anesthesiologists. All nonemergent intubations are premedicated with atropine, fentanyl, and rocuronium prior to intubation. Indications for intubation vary among individual providers but commonly include respiratory failure, elevated FiO₂ requirement, respiratory acidosis, surfactant administration, apnea, increased work of breathing, or elective procedures.

Inclusion criteria were intubations occurring in the NICU between January 2017 and June 2019 with a completed quality data form. Exclusion criteria were delivery room intubations, since quality data is not collected on these intubations, patients with anatomic airway abnormalities, and any intubations without a completed quality data form or with incomplete data regarding complications. Emergent intubations were not premedicated and typically did not either have complications recorded or quality data submitted due to the procedures for such data collection. Additional chart review was completed on all patients.

The primary outcome was the presence of one or more complications during neonatal intubation; defined per the quality data sheet checklist to include bradycardia (heart rate of <60 beats per minute for ≥5 seconds), desaturation (pulse oximeter <60%), airway trauma, difficulty with bag-mask ventilation, emesis, chest wall rigidity, pneumothorax, chest compressions, or code medications. For the purposes of this paper, both adverse events and vital sign abnormalities were included in the definition of complications. These criteria were modified from the study by Hatch et al⁷ and were simplified to streamline data collection in real time by the recorder during the intubation. Other relevant data collected in real time during intubations included: date and time of intubation, dose and administration of premedication, number of attempts, GA at birth, PMA, day of life, and current weight. The quality checklist filled out at the time of intubation is completed by the medical team at the time of intubation. Typically, the bedside nurse completes most of the data in real time during the intubation, and complications are reviewed with the team leader immediately after the intubation and recorded in the EMR. Although the recorder initially fills out the form, the team leader is required to sign off on the form and confirm the recorded data at the conclusion of the intubation event. Discrepancies between the quality data and the EMR are resolved with the team leader when the quality data is entered into the database, typically within 1 month of the intubation event. Retrospective chart review was performed on all patients to collect patient characteristics and variables including confirmation of variables collected at the time of intubation. Additional variables collected included: role of first intubation provider (neonatal-perinatal medicine fellow, neonatal nurse practitioner, pediatric resident, neonatal-perinatal attending physician, respiratory therapist, or other), partial pressure of oxygen and carbon dioxide on blood gas results in the 12 hours prior to intubation, and FiO₂ at baseline one hour prior to intubation. Primary diagnoses and indication for intubation were abstracted from the medical record and included any of the following: pulmonary hypertension, respiratory distress syndrome (RDS), surfactant administration, transient tachypnea of the newborn (TTN), necrotizing enterocolitis (NEC), sepsis, chronic respiratory failure, acute respiratory failure, acute on chronic respiratory failure, apnea or bradycardia, reintubation after unplanned extubation, planned procedure, air leak (ie pneumothorax), upper airway obstruction, airway and/or craniofacial anomaly, neurologic impairment, or congenital heart disease. Diagnoses and indication for intubation were determined based on the history and physical, daily progress note, procedure note, or clinical event note in the days before and after the intubation encounter. Diagnoses or indications for intubation were included if they were mentioned in any of these notes. For example, a patient undergoing evaluation for sepsis as a possible differential diagnosis for respiratory distress would have sepsis included as one of the diagnoses.

Power analysis:

We estimated that with a sample size of approximately 470 patients and a baseline complication rate of about 40%, with an alpha of 0.05, and a power of 0.8, the detectable difference in complication rates between groups would be approximately 15%.

Statistical Analysis:

Demographics and clinical characteristics of the patients were described using median and interquartile range for continuous variables; and frequency and percentages for categorical variables, as appropriate. Categorical variables were analyzed with the chi-square test or Fisher Exact test; continuous variables were assessed by Mann Whitney U test. Potential covariates of interest included FiO₂, PCO₂, GA at birth, PMA, and weight. For continuous variables of interest, a threshold analysis was performed to identify the optimal cut-point at which complications increase, in order to provide relevant data for clinical use. Potential covariates that differed between groups with a p-value of <0.2 in the initial analysis were used in the generalized estimating equations (GEE). Additionally, administration of rocuronium was included in the model as a known factor which may decrease the risk of adverse events. GEE were used to account for correlation among outcomes for infants who experienced multiple intubation encounters. We carried out two GEE, the first treating the predictors as continuous and the second categorizing them based on the optimal cut-point. A p-value less than 0.05 or the odds ratio confidence intervals not including 1 were considered statistically significant for the analysis. Variables were only used if they would be known prior to intubation in order to identify predictors of complications. Patients with missing data on complications were excluded from the analysis. Patients with other missing variables such as FiO₂ or CO₂ were included in the analyses for data which were available. All the analyses were performed using SAS software, version 9.4 (SAS Institute, Cary, NC) and R software, version 4.0.0, Vienna, Austria (packages used: OptimalCutPoints, reportROC, geepack). Study data were collected and managed using Research Electronic Data Capture Tools (REDCap) hosted at University Hospitals⁹.

Results:

473 intubation encounters occurred between January 28, 2017 and June 27, 2019 with available quality data. Six subjects did not have complications recorded at the time of intubation and were excluded leaving 467 intubation encounters in 352 infants in the analysis. Nearly all intubation encounters were non-emergent and pre-medicated; 13 (2.8%) were emergent, 14 (3.0%) did not receive fentanyl, and 44 (9.4%) did not receive rocuronium. 41.5% (N=194) intubation encounters had the presence of 1 or more complications and 58.5% (N=273) had no complications. The three most common complications were desaturation <60%, difficulty with bag-mask ventilation, and bradycardia <60 for ≥5 seconds (E-table 1). We found that almost all complications were also accompanied by a desaturation <60%, which the recorder could easily document during the intubation encounter. There were 7 intubation encounters with complications which did not include desaturation <60%; of these 1 had airway trauma, 4 had difficulty with bag-mask ventilation, and 2 had other complications. Of the encounters with desaturation during intubation, 108 included a desaturation alone. Of those with multiple complications, 51 had desaturation plus one additional complication (34 with difficulty with bag-mask ventilation, 12 with bradycardia, 3 with airway trauma, 2 with emesis); 23 had desaturation plus two additional complications (22 with difficulty with bag-mask ventilation, 13 with bradycardia, 6 with airway trauma, 5 with chest wall rigidity); and 5 had desaturation plus 3 additional complications (5 with bradycardia, 4 with difficulty with bag-mask ventilation, 2 with emesis, 1 with pneumothorax, 1 with chest compressions, 1 with code medications, 1 with other). The most common diagnoses and indications for intubation were acute respiratory failure (75.4%), respiratory distress syndrome (67.2%), and sepsis (49.0%) (E-table 2). Median number of attempts per intubation encounter was 2 (IQR 1-3). Subjects with complications compared to those without complications had a median GA at birth of 31 weeks compared to 29 weeks (p<0.001), PMA of 33 weeks compared to 32 weeks (p=0.352), weight of 1970g compared to 1540g (p=0.091), chronological age of 1 day compared to 3 days (p=0.001), and FiO₂ of 0.45 compared to 0.50 (p=0.183). Median chronological age overall was 2 days (range 0-187). Median pCO₂ was 59 for both groups (p=0.716). Subjects without complications had fewer attempts with 59.3% compared to 20.6% of subjects with complications having only 1 attempt (p-value <0.001) (Table 1).

Table 1:

Patient Characteristics and Complications

	No Complications (N=273)	Complications (N=194)	P-value
Fraction of Inspired Oxygen (median [IQR])	0.45 [30, 65]	0.50 [32, 75]	0.183a
Gestational Age at Birth (whole weeks) (median [IQR])	31 [26, 36]	29 [25, 35]	<0.001a
Postmenstrual Age (whole weeks) (median [IQR])	33 [29, 37]	32 [28, 37]	0.352 a
Weight at the time of Intubation (grams) (median [IQR])	1970 [1110, 2885]	1540 [1032, 2728]	0.091 a
Chronological Age (days) (median [IQR])	1 [0, 16]	3 [0, 37]	0.001 a
pCO2 (median [IQR])	59 [51, 69]	59 [50, 70]	0.716 a
pO2 (median [IQR])	66 [50, 90]	67 [53, 82]	0.631 a
Administration of Rocuronium (n, %)	253 (92.7)	170 (87.6)	0.066 b
Role of first airway provider (n, %)
Attending Neonatologist	4 (1.5)	3 (1.5)
Fellow	104 (38.1)	61 (31.4)
NNP (or NNP Student)	58 (21.2)	56 (28.9)
Other	6 (2.2)	3 (1.5)
Resident	99 (36.3)	71 (36.6)
Respiratory Therapist	2 (0.7)	0 (0.0)	0.347b
Number of Attempts (n, %)
1 attempt	162 (59.3)	40 (20.6)
2 attempts	56 (20.5)	81 (41.8)
3 attempts	35 (12.8)	43 (22.2)
4 or more attempts	20 (7.3)	30 (15.5)	<0.001b

Abbreviations:

^a:Mann Whitney U test;

^b:Chi-square test

Variables entered into the GEE with both continuous and binary variables included administration of rocuronium, chronological age, FiO₂, and weight. GA at birth was excluded due to issues with collinearity with weight. Weight was chosen as a more clinically meaningful characteristic throughout the hospital stay. Threshold analysis was done to determine ideal cut points to classify continuous variables as binary variables, which revealed a chronological age of 21.5 days (AUC 0.585), FiO₂ 0.605 (AUC 0.537), and weight 1565g (AUC 0.454) (E-table 3, E-figure 1). GEE with binary variables revealed that infants <21.5 days in chronological age (OR 0.45 CI 0.30-0.69) had decreased odds of complications. Infants <1565g in weight (OR 1.52 CI 1.04-2.23) had increased odds (Table 2).

Table 2:

Generalized Estimating Equation of Patient Factors Associated with Odds of Complications with Binary Variables

Characteristic	Odds Ratio	95% Confidence Interval
Chronological Age ≤21.5 days	0.45	0.30; 0.69
Fraction of Inspired Oxygen ≤0.605	0.73	0.48; 1.12
Weight ≤1565g	1.52	1.04; 2.23
Rocuronium Administered	0.59	0.31; 1.11

Variables are categorized based on the optimal cut-off points obtained from the ROC analysis

The GEE model is adjusted for Age, Fraction of Inspired Oxygen, Weight, and Administration of Rocuronium

In the GEE where predictors were treated as continuous variables, with every 1 week increase in chronological age (OR 1.07, CI 1.03-1.12), we expect to see a 7% increase in the odds of complications; and for every 500 gram increase in weight (OR 0.89 CI 0.81, 0.98), we expect to see an 11% decrease in the odds of complications. (Table 3). Although number of attempts differed between groups with a p-value of <0.001, this was not included in the GEE as this is not a factor that would be known prior to intubation.

Table 3:

Generalized Estimating Equation Analysis of Patient Factors Associated with Odds of Complications with Continuous Variables

Characteristic	Odds Ratio	95% Confidence Interval
Chronological Age	1.07	1.03; 1.12
Fraction of Inspired Oxygen	1.06	0.98; 1.14
Weight	0.89	0.81; 0.98
Rocuronium Administered	0.56	0.29; 1.06

A unit difference is as follows: Age (7 days); Fraction of Inspired Oxygen (0.1); Weight (500 grams)

The GEE model is adjusted for Age, Fraction of Inspired Oxygen, and Weight

Due to the fact that prior studies have examined desaturation separately from other adverse events, post-hoc analysis was performed using GEE looking at the odds of desaturation <60% alone rather than all complications. These results showed a decreased odds of desaturation with chronological age ≤21.5 days (OR 0.48, CI 0.31-0.73) or weight ≤1565g (OR 1.47, CI 1.00-2.17) (E-table 4). Additionally, when the predictors were treated as continuous variables, with every 1 week increase in chronological age (OR 1.07, CI 1.03-1.12), we expect to see a 7% increase in the odds of desaturation, and for every 500 gram increase in weight (OR 0.90, CI 0.82-0.99), we expect to see a 10% decrease in the odds of desaturation (E-Table 5).

Missing data included the following: CO₂ level for 17% (N=80), PO₂ for 66% (N=309), and FiO₂ for 6.5% (N=31); when blood gas was not performed within 12 hours prior to intubation, blood gas was not an arterial sample, or FiO₂ was not recorded in the EMR within 30 to 90 minutes prior to intubation, respectively.

Discussion:

This large single-center retrospective cohort study is the first study to specifically focus on describing patient characteristics associated with complications, including adverse events, desaturation, and bradycardia, in neonatal intubations. This analysis identified two characteristics that can be used to identify patients at higher risk of complications during intubation: weight and chronological age. Although these risk factors are not modifiable, identification of high-risk patients may allow for focused interventions to reduce risk for those patients.

Complications in neonatal intubations have recently been described in more detail in the literature^{1; 2} including the creation of a large multicenter database (NEAR4NEOS) to further evaluate neonatal intubations and adverse events⁶. It is clear that complications; including both adverse events and vital sign abnormalities, are occurring at a high rate in neonatal intubations. This institution found a similar rate of desaturation as found in other studies, with variation expected due to center differences as well as differences in the definition of desaturation^{1; 2; 6}. Foglia et al described a rate of 22% of adverse tracheal intubation associated events and 51.1% rate of desaturation 20% below baseline in a single center¹. Foglia et al then described a rate of 18% of adverse tracheal intubation associated events and 48% desaturation rate 20% below baseline in NICU intubations in the multicenter NEAR4NEOS registry⁶. Hatch et al described a rate of 39% of adverse events with 44% of desaturation <60% in a single center². Desaturation <60% was included in the definition of complications in this cohort; 187 of 194 (96.4%) of intubation encounters with a complication and 40% of 467 total encounters included desaturation <60%. The definition of complications was simplified in this report to include both adverse events and vital sign abnormalities. This change allowed for more rapid recording of complications after each intubation encounter, which makes the use of this checklist more easily generalizable to NICUs without extensive support structures for collection of quality data. Including desaturation and bradycardia as part of the definition allows for objective physiologic definitions of complications which can be recorded by any observer during the intubation encounter. This change makes these results difficult to compare to previous studies and may explain the differences described. Hatch et al described rates of adverse events and modifiable factors contributing to those events, and did not see a difference in postmenstrual age, postnatal age, or weight between patients with and without adverse events². This may be due to differences in population with most emergent intubations not included in this report. This study did not show any difference in CO₂ level between groups with and without complications, in contrast to the study by Foglia et al which showed that rising CO₂ was a more common indication for intubation when adverse events were present¹. This may be explained by objective examination of blood gas data compared to a subjective assignment of ventilation failure as the primary indication for intubation in that population¹. This study did not show a difference in complications based on role of the first airway provider. This may be due to lack of detail regarding experience level of the first airway provider, such as level of training or number of intubations performed, which has been shown to be associated with adverse events in multiple other studies^{1; 6}. Another contributor may be the general practice in this NICU, where typically the least experienced provider attempts first.

Chronological age was a significant factor which increased the odds of complications with increasing age. The authors hypothesize that infants who are getting intubated at an older age have more severe lung disease, which is persistent and chronic, as opposed to infants intubated in the first days of life for indications such as surfactant administration. These younger infants therefore may have less severe lung disease and more reserve which places them at decreased risk of complications during intubation.

Potential strategies for reducing complications include those shown in prior studies: intubation by an attending physician compared to a resident physician¹, use of paralytic medication^{1; 6}, use of videolaryngoscopy⁶, reducing number of attempts², and reducing emergent intubations². Hatch et al described a decreased incidence of adverse events in providers who have completed >10 successful intubations with an OR of 0.7 and further improvement with practitioners who have completed >40 successful intubations with an OR of 0.47 (p-value 0.09)². In the current educational climate, pediatric residents have fewer experiences with intubation during residency in both neonates and pediatric patients^{10, 11}, which is a trend which has likely continued due to changes in both clinical care of infants with respiratory distress^12–19, as well as changes in pediatric residency training.²⁰ The high rate of complications during neonatal intubations stands in contrast to the continued desire to train pediatric residents to intubate. Neonates have increased risk of desaturation, commonly with a more profound drop in oxygen saturations and a longer duration to recover compared to adults²¹ and pediatric patients²². Duration of an intubation attempt is recommended to be less than 30 seconds in neonates¹², but it is uncommon for trainees to successfully intubate in less than 30 seconds²³. However, the American College of Graduate Medical Education (ACGME) states that pediatric residents must demonstrate procedural competence by performing neonatal endotracheal intubation²⁰. Therefore, identifying patients at lower risk of complications can be used to continue educational efforts and intubation opportunities for pediatric trainees while minimizing risk to the patient. Additionally, premedication including paralysis for intubation has been recommended by the American Academy of Pediatrics since 2010²⁴. However many NICUs continue to intubate without premedication or with sedation alone²⁵. Although not statistically significant; in this study administration of rocuronium resulted in a 41% decreased odds of complications (OR 0.59 CI 0.31-1.11) (Table 2). Technology such as laryngeal mask airway as an alternative airway or videolaryngoscopy, which has been shown to decrease adverse events and increase success^{6; 26}, are lacking for the smallest infants, but will be a useful strategy when available for this population.

One strength of this study includes identification of objective patient characteristics which can be used across centers as markers for increased risk of complications during intubation. Another strength is the prospectively collected data on complications, minimizing recall bias for the primary outcome as they are recorded at the time of intubation. The authors acknowledge there may be some inaccuracy based on short-term recall, situational awareness, and reliance on the airway team to identify complications, but the authors believe this was minimal and consistent throughout the project. Limitations include the lack of data on delivery room intubations and emergent intubations due to the nature of the data collection for quality data at this institution. We were unable to determine the total number of intubations occurring at this center outside of those with quality data forms submitted. The intubations which are not recorded are more likely to be in unstable patients and high-stress situations, therefore likely have a higher rate of complications. Therefore, these results may underestimate the true incidence of complications in this population. Additionally, this study represents a population with nearly universal use of premedication prior to intubation, as recommended by the American Academy of Pediatrics²⁴. These factors may limit the generalizability of these results. The complications recorded in this study are simplified from prior studies for ease of use and efficient documentation in real time by any observer during the intubation encounter. This change in the definition and the inclusion of desaturation <60% with other complications may limit the ability to compare our results with other similar published work. The authors did note that with this simplified definition almost all intubations with a complication included desaturation <60%. However, of the patients with a desaturation (n=187), the majority experienced a desaturation alone without any other complication noted (n=108). The definition of complications in this study excluded some of the more common adverse events described by other authors including esophageal intubation, mainstem intubation, laryngospasm, hypotension, pain, or gum, dental, or lip trauma. Some of these adverse events may be accompanied by vital sign changes such as desaturation or bradycardia. Future study is needed to establish a universal definition of complications which are clinically meaningful and easy to record at the bedside.

This report adds another center’s experience with complications in neonatal intubation and further serves to highlight the need for systematic observation of complications during neonatal intubation and quality improvement work to address modifiable risk factors⁷, especially in patients who are known to be high-risk prior to intubation.