Results
Altogether, 20 health care providers including Neonatal Fellow/Neonatal attending physicians (40%), postgraduate medical trainees (25%) and senior nursing staff (35%) performed POCUS on neonates requiring intubation. All neonatal health care providers were trained and certified for airway POCUS (Table 1).
A total of 348 neonates were enrolled in the study from July 2018 to June 2019. The mean gestational age of newborns was 32.6 (SD± 4.8) weeks. More than half (58%) of intubations were in an emergency scenario while others were elective for procedures under anesthesia. The major reasons of emergent intubations included hemodynamic compromise (31.2%), respiratory failure (29.7%) and administration of surfactant (20.8%). The most common medical diagnosis of enrolled participants was respiratory distress syndrome, which accounted for 199 (57.2%) neonates. The mortality among the entire cohort was 23.6% (Table 2).
In total, 318 (91.3%) were tracheal intubations while 30 (8.7%) were esophageal intubation as proven by at least two standard methods (auscultation, capnography, and chest radiography). Moreover, 266 (76.4%) intubations were confirmed by all three methods. X-ray was not available for 82 (23.6%) intubations.
We evaluated the diagnostic accuracy of POCUS user interpretation compared with at least two standard-of-care methods. Altogether, 344 POCUS user interpretations were consistent with standard of care; however, three were reported as false positive and one as false negative. Therefore, the sensitivity and specificity were 99.7% and 91%, respectively, with 98.85% agreement (Kappa: 0.93; p < 0.0001).
The median time required for POCUS interpretation was 3.0 seconds (IQR: 3.0-4.0) with tracheal intubation as 3.0 seconds (IQR: 3.0-4.0) and esophageal intubations as 4.5 seconds (IQR: 3.0-8.0). Similarly, the recorded time measured for auscultation and capnography was 6.0 seconds (IQR: 5.0-7.0) and 3.0 seconds (IQR: 3.0-4.0), respectively. When applying the Wilcoxon signed-rank test to the time required for gold standard methods, auscultation and X-ray times were significantly higher than to POCUS interpretation time. However, CO2 detector time (p-value: 0.594) was similar to POCUS interpretation time (Table 4).
We also compared the interpretation of POCUS images by user in the NICU and the POCUS expert and found the same interpretation as the expert assessment (97.7% images). Only two images were inconsistent between user and expert interpretations. This analysis yielded 94% specificity and 100% sensitivity. Furthermore, there was 99.4% agreement (Kappa: 0.96; p < 0.0001) between the POCUS user and expert.