Introduction: Neonatal tracheal intubation (TI) outcomes have been assessed by role, but training level may impact TI success and safety. Effect of physician training level (PTL) on the first-attempt success, adverse TI-associated events (TIAEs), and oxygen desaturation was assessed. Methods: Prospective cohort study in 11 international NEAR4NEOS sites between October 2014 and December 2017. Primary TIs performed by pediatric/neonatal physicians were included. Univariable analysis evaluated association between PTL, patient/practice characteristics, and outcomes. Multivariable analysis with generalized estimating equation assessed for independent association between PTL and outcomes (first-attempt success, TIAEs, and oxygen desaturation ≥20%; attending as reference). Results: Of 2,608 primary TIs, 1,298 were first attempted by pediatric/neonatal physicians. PTL was associated with patient age, weight, comorbidities, TI indication, difficult airway history, premedication, and device. First-attempt success rate differed across PTL (resident 23%, fellow 53%, and attending 60%; p < 0.001). There was no statistically significant difference in TIAEs (resident 22%, fellow 20%, and attending 25%; p = 0.34). Desaturation occurred more frequently with residents (60%), compared to fellows and attendings (46 and 53%; p < 0.001). In multivariable analysis, adjusted odds ratio of the first-attempt success was 0.18 (95% CI: 0.11–0.30) for residents and 0.80 (95% CI: 0.51–1.24) for fellows. PTL was not independently associated with adjusted odds of TIAEs or severe oxygen desaturation. Conclusion: Higher PTL was associated with increased first-attempt success but not TIAE/oxygen desaturation. Identifying strategies to decrease adverse events during neonatal TI remains critical.
Proficiency in neonatal tracheal intubation (TI) is critically important for providers in delivery rooms and neonatal intensive care units (NICUs). The suboptimal TI technique has been associated with direct airway injury, patient deterioration, intraventricular hemorrhage [1, 2], neurodevelopmental impairment , and death [4-6].
Reported neonatal TI success rates range from 20 to 60% [7-9]. Pediatric resident TI competency is particularly concerning, as success rates appear to be declining [7-13]. Numerous factors likely contribute, including limited work hours and NICU rotations , increased presence of advanced pediatric providers , and clinical practice changes, including care of infants born through meconium-stained amniotic fluid  and utilization of noninvasive ventilation strategies [17, 18]. There is great concern that changes in the practice environment have negatively impacted current physician trainees’ neonatal TI competency.
Physicians’ TI competency is critical to avoid short- and long-term adverse events and ensure patient safety. Considering that the current practice environment offers limited TI opportunities, there may be increased risk to patients intubated by physician trainees if not selected deliberately and supervised closely. Additionally, many residents may not be competent to safely perform TI after graduation, although this skill remains important for pediatricians [19, 20] in rural settings [21, 22] or entering critical care subspecialty fellowships .
Since most studies classify providers according to role, a knowledge gap exists in understanding how trainee TI skills improve throughout residency (postgraduate year [PGY] 1–3) and fellowship (PGY 4–6), specifically related to success rates and outcomes. The aim of the present study was to evaluate TI performed by physicians across the continuum of pediatric postgraduate medical training (including each specific year of residency/fellowship training) and attending physicians to assess for differences in success rates, adverse TI-associated events (TIAEs), and severe oxygen desaturation (oxygen desaturation ≥20%). We hypothesized that intubations performed by junior trainees would have lower success rates and higher rates of adverse outcomes.
Setting and Design
This retrospective cohort study was conducted using prospectively collected data from 11 international NICUs from North America, Europe, and Asia participating in the National Emergency Airway Registry for Neonates (NEAR4NEOS) quality improvement registry between October 2014 and December 2017 [24-28]. All NEAR4NEOS sites were granted Institutional Review Board approval or deemed exempt as quality improvement initiatives. Informed consent was waived. Primary TIs performed by pediatric/neonatal physicians across the physician training spectrum (residents, neonatal-perinatal medicine [NPM] fellows, and neonatology attendings) were included. Intubations performed outside the NICU (i.e., delivery room), performed for tube exchanges, and performed by nonphysician clinicians (including physician assistants, nurse practitioners, and respiratory therapists) and non-NICU-based clinicians (including pediatric surgeons, otolaryngologists, and anesthesiologists) were excluded from this analysis. Participating NICUs did not have a protocol to specify the intubating clinician’s discipline or training level; this selection was left to the discretion of the bedside team.
Study Exposure and Outcomes
The exposure of interest was the training level of the physician performing the initial TI attempt.
The primary outcome was the first-attempt success rate. Secondary outcomes included adverse TIAEs and severe oxygen desaturation.
Operational definitions utilized in this study were developed a priori in NEAR4NEOS, as previously described . In brief, a course refers to a single approach to airway management, including device and premedication. An intubation attempt begins with device (i.e., laryngoscope) placement into the patient’s oropharynx and ends with either device removal or tracheal tube placement. A successful course of airway management was defined as an endotracheal tube placed into the trachea, with confirmation by chest rise, auscultation, carbon dioxide detection, laryngoscopic assessment (using video laryngoscopy [VL] or secondary independent laryngoscopy), and/or chest radiography. First-attempt success was defined as a successful intubation by the first provider in the first attempt of the first course.
Safety outcomes included adverse TIAEs and severe oxygen desaturations. Adverse TIAEs were categorized into severe and nonsevere, as previously described . Severe oxygen desaturation was designated as ≥20% decrease in oxygen saturation from the highest level immediately before the procedure to the lowest oxygen saturation during the course.
Analyses were conducted using Stata 15.0 (Stata Corp, College Station, TX, USA). Univariable analysis was performed to assess for association between physician training level (PTL) and patient factors, practice factors, and outcomes using the χ2 test or Fischer’s exact test for categorical variables and the Wilcoxon rank-sum test for nonparametric variables. Multivariable analysis was performed with generalized estimating equation to assess for independent association between PTL and TI outcomes while controlling for potential confounders. All analyses were a priori specified and approved by the NEAR4NEOS manuscript oversight committee. p value <0.05 was considered statistically significant.
Patient and Practice Characteristics
Of 2,608 qualifying primary TIs, 1,298 (49.8%) were first attempted by pediatric/neonatal physicians. Initial TI attempts were performed most often by fellows (61.6%), followed by residents (30.1%), Table 1. PTL was significantly associated with patient age and weight at TI. The gestational age among infants intubated by fellows was significantly lower (median 27 weeks, IQR: 25–34 weeks) than in infants intubated by residents or attendings (p < 0.001). Residents often performed TI for ventilation failure (41%), surfactant administration (14%), or electively for a planned procedure (28%). Conversely, attending physicians were more likely to perform TI in infants with a history of difficult airway (DA) (26%), airway/craniofacial anomaly (14%), or congenital heart disease (14%). TIs performed by attending physicians more frequently had indications of unstable hemodynamics (7%) or unplanned extubation (23%).
Table 2 details practice-related factors. VL was similarly utilized across PTL (22% in residents, 24% in fellows, and 27% in attendings, p = 0.419). Sedation and paralysis were utilized in slightly less than half of attempts across provider groups. Attendings were most likely to perform TI without sedation or paralysis (24% in residents, 34% in fellows, and 43% in attending, p < 0.001).
First-attempt success rates were significantly different among PTL: 23% in residents, 53% in fellows, and 60% in attendings; p < 0.001, Table 3. First-attempt success rates gradually increased with PTL: 16% in PGY-1 residents, 19% in PGY-2 residents, 34% in PGY-3 residents, 51% in PGY-4 fellows, 55% in PGY-5 fellows, and 60% in attendings; p < 0.001, nonparametric χ2 for trends, Figure 1. Among provider groups, there was no difference in the frequency of any TIAEs (22% in residents, 20% in fellows, and 25% in attendings; p = 0.344) or severe TIAEs (5% in residents, 5% in fellows, and 8% in attendings; p = 0.354) (Fig. 1; online suppl. Table A; see www.karger.com/doi/10.1159/000516372 for all online suppl. material). Online suppl. Table B demonstrates a breakdown for each TIAE. Severe oxygen desaturation occurred more frequently with residents (65%), compared to fellows (52%) and attendings (58%); p < 0.001.
Table 4 summarizes results after controlling for patient and practice factors associated with provider levels in the multivariable analysis. The adjusted odds ratio (OR) for first-attempt success for residents was 0.18 (95% CI: 0.11–0.30; p < 0.001) and for fellows was 0.80 (95% CI: 0.51–1.24; p = 0.328) using attending performance as reference.
PTL was not independently associated with occurrence of any TIAE (adjusted OR for residents: 0.91, 95% CI: 0.52–1.59, p = 0.732; adjusted OR for fellows: 0.69, 95% CI: 0.41–1.16, p = 0.163, with attending as reference; online suppl. Table C) nor with rates of severe oxygen desaturation (adjusted OR for residents: 1.56, 95% CI: 0.96–2.53, p = 0.073; adjusted OR for fellows: 0.82, 95% CI: 0.52–1.28, p = 0.384, with attending as reference; online suppl. Table D).
This retrospective cohort study demonstrated significant differences in first-attempt success rates in neonatal TI across PTL (resident, fellow, and attending) and provides granular information on incremental improvement with each physician training year. Despite improved first-attempt success rates, there was no significant difference in secondary outcomes, including occurrence of adverse TIAE or severe oxygen desaturation.
To our knowledge, this is the first multicenter study to evaluate neonatal TI success and safety outcomes using discrete 1-year increments in pediatric physicians across residency, NPM fellowship, and ultimately as attending neonatologists . These findings are important to elucidate due to tremendous learning curves encountered early in a postgraduate training program, as well as the need to evaluate performance more precisely for summative assessment of trainees preparing for the next step in training, or entry into unsupervised practice. Differences in TI success and safety across PTL may impact individual provider’s educational opportunities, highlight the need for alternative airway management strategies (such as laryngeal mask airways), and inform development of observational assessment tools for TI training . These findings may aid supervising educators in selecting appropriate TI attempts for trainees at a given level, considering factors such as patient size, indications for intubation, existing comorbidities, and hemodynamic stability. Supervisors should also optimize intubation conditions, such as use of premedication when appropriate.
We observed improvement in first-attempt success rate across the continuum of physician postgraduate training. Compared to attending neonatologists, residents had significantly lower first-attempt success rates, especially earlier in their training. These findings are consistent with findings in pediatric ICUs  and may impact discussions on whether it is feasible for all graduating pediatric residents to demonstrate competency in TI. This situation has been challenged by decreased TI opportunities due to practice changes, such as more conservative management of infants born through meconium-stained amniotic fluid , widespread use of noninvasive ventilation in the neonatal/pediatric ICU settings [8, 11, 31], and increasing utilization of advanced pediatric providers . This raises concerns for technical proficiency of new NPM fellows at the beginning of training.
Our attending success rates are lower than what might be considered ideal. Attendings, in this study, were often intubating patients with challenging anatomy or unstable hemodynamics suggesting stressful clinical situations, or where other providers had failed. We hypothesize that attending success rates would be much higher if attendings were intubating on the first attempt in patients with less challenging characteristics and more stable physiologic status, allowing better optimization of intubating conditions. These factors may explain differences in success rates between our study and operating room (OR) intubations . Another possibility for this suboptimal success rate is that current-day attendings have had fewer opportunities to refine their intubation skills during training than their predecessors [8, 11, 15, 16, 31], coupled with infrequent opportunities to refresh their skills clinically while supervising their team (given the decreased procedural frequency overall, as well as the need for trainees and other providers to achieve and maintain proficiency) . As such, it remains critically important to identify robust assessment methods to ensure fellows are competent in intubation prior to completion of training (ideally through observational assessment of psychomotor skills, or a learning curve analysis, such as cumulative sum [33-36]) and to provide ongoing procedural exposure for experienced intubators.
It is notable that there was no significant association between PTL and occurrence of adverse TIAE or severe oxygen desaturations in the multivariable analysis. We hypothesize that this may be due to close supervision of junior trainees in the NICU, with intubation attempts only permitted in patients with hemodynamic stability and less likelihood of DA anatomy. Conversely, the lack in difference in adverse safety outcomes across PTL may also be related to increased physiological instability and anatomic difficulty in the patients being intubated by attendings. While these patient factors were adjusted as covariates in the multivariable analysis, it is likely that there were unmeasured confounders associated with higher rates of adverse TIAEs and preference for a physician of higher training level to perform TI in a given situation. These factors may also be reflected in the “U-shaped” pattern of severe oxygen desaturation outcomes (with higher rates of desaturation noted in both novice and experienced intubators). Junior providers may have required additional time to complete the procedure while experienced providers may have encountered patient instability or anatomic challenges.
In our multivariable model, several factors were associated with higher first-attempt success rates. Factors such as use of sedation/paralysis and indication for TI procedure make sense with regard to facilitating first-attempt success ; however, TI indication for upper airway obstruction and unplanned extubation were associated with higher first-attempt success rates without clear reasons. We speculate these indications may be potentially associated with more stable patient status and provide suitable intubation conditions.
Conversely, additional factors which may be suspected to confer decreased first-attempt success rate, including history or examination suggestive of a DA, did not demonstrate statistical significance in this study. This may be a dilutional effect due to the overall relatively low first-attempt success across TIs, with or without DA. Sawyer et al.  published an analysis of NEAR4NEOS data where “difficult intubation” was defined specifically as an intubation requiring 3 or more attempts by a fellow/attending. Not surprisingly, patients with difficult intubation had higher risk for TIAE (OR 4.9) and severe oxygen desaturation (OR 4.2). In our study, while it was unusual for attending physicians to perform initial intubation attempts (n = 108 out of 1,298 courses), approximately 25% of these were performed in infants with either history or clinical exam suggestive of DA.
Notably, use of VL in this analysis was not associated with a significant difference in TI outcomes. Previous authors have demonstrated improvement in clinical TI first-attempt success rates (although without a reduction in duration, number of required attempts, or adverse events) with VL, particularly for junior trainees [24, 37-40]. Additionally, a recently published randomized control study demonstrated that use of VL in infants was associated with higher first-attempt success in the OR . Our multivariable analysis, however, indicated that there was no significant association between VL use and first-attempt success in this cohort. This difference may be due to the fact that our patients are much smaller than those in the OR study, or that intubation conditions may be different (e.g., a TI occurring in the OR or selected for a trainee to perform may be more controlled), contributing to higher success rates in other settings.
There are limitations to this study. First, data on intubation success rates and adverse outcomes were collected by self-report. Despite extensive training on operational definitions at each site and rigorous local protocols to capture data on all TI courses, it is possible that reporting bias exists. Most institutions participating in NEAR4NEOS are academic centers with residency and fellowship training programs. However, results may not be representative of all centers with respect to local practice or physician training structure. Additionally, trainee TI proficiency may differ depending on presence of other trainees or clinicians (i.e., pediatric residents may have more opportunities for TI in institutions without a neonatology fellowship). We did not collect data on existing center-specific educational curricula for TI, which may impact success and safety outcomes. Importantly, our study did not follow individual physician trainees over time, and the analyses did not consider individual learning curves. A different type of learning curve analysis such as cumulative sum would be appropriate for such an analysis [33-36]. Finally, we do not have data on duration of intubation attempts, which may significantly impact first-attempt success rates, adverse TIAEs, and oxygen desaturation. Future studies using video recording carts will facilitate such data collection [41, 42].
First-attempt success rates increased with advancing PTL, but training level was not independently associated with TIAE or severe oxygen desaturation. Future efforts should focus on identifying strategies to improve intubation safety, having experienced providers intubate high-risk babies, and by examining methods to avoid adverse events and desaturation.
Statement of Ethics
The Children’s Hospital of Philadelphia served as the reviewing IRB for this multicenter study (IRB No. 09-007253). Additionally, NEAR4NEOS was granted IRB approval or was deemed exempt from IRB oversight at each individual center. All sites granted waiver of informed parental consent for data collection and analysis.
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
This project was supported by the National Institutes of Child Health and Human Development (NICHD; R21HD089151-01A1). A.N. is supported by the NICHD (R21HD089151-01A1). E.F. is supported by an NICHD Career Development Award (K23HD084727).
L.J. conceptualized and designed the study, analyzed the data, drafted the initial manuscript, and reviewed and revised the manuscript. All authors except J.S. coordinated and supervised data collection. J.S. and A.N. performed statistical analyses. All authors interpreted the data and critically reviewed and revised the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
Elizabeth E. Foglia and Akira Nishisaki are the colast authors.