I think I'm just going to go ahead and start. I don't know what you guys say, since we all have an hour. I'm Niranjan Vijayakumar. I'm a cardiac intensivist at Boston Children's Hospital. So I'm just going to go ahead and kickstart the session, I guess. I am extremely fortunate and proud to be presenting this on behalf of my team from Boston Children's Hospital. I'm going to be talking about our work titled Ultrasound Screening for Vocal Cord Dysfunction After Congenital Heart Surgery in Children. And we don't have any financial conflicts of interest to disclose. Over the next 10, 12 minutes or so, I'm just going to quickly talk about the background of why we went into this project, our main objectives, the methodology that we engaged in, and give a brief overview of our results. And then I want to spend a couple of minutes talking about our clinical impressions from doing this project about use of the vocal cord ultrasounds in screening for these children. And then present our conclusions. And then spend maybe a minute or so talking about our future directions on this project. So to go into the background, as we all know, we are moving away from looking at mortality as the only outcome in management of post-op congenital heart surgery. And one of the important things that has come up is feeding dysfunction. And it's a very common post-operative morbidity in these children. And it has been associated with very high resource utilization. The etiology of feeding dysfunction has been shown to be multifactorial. And it is a complex mechanism that goes into the suck-swallow reflex. And a major portion of that has also been promoted by vocal cord dysfunction. And earlier identification of this vocal cord dysfunction, vocal fold mobility issues, can promote a reduction in resource utilization. A couple of articles that I wanted to highlight. This one from 2017 shows that the total hospital cost in dollars was about $170,000 in patients who had vocal fold mobility issues, when compared to about $120,000 for patients without vocal fold mobility issues. And a predominant driver for that is readmission for feeding difficulty for these patients. And also the dependence on asentric or surgical tube feeding. Going more into vocal fold mobility issues, this is probably very familiar for most of you here. But it's primarily driven by stress or injury to the rectal laryngeal nerve. And it has got to do with the anatomic position of the rectal laryngeal nerve around the aortic arch. And therefore, the incidence is pretty high for procedures that concern the arch. It is a significant part of the swallow-suck reflex, like I mentioned before. And the glottic approximation is a big driver for laryngeal penetration during the swallow reflex. And the current gold standard for identification of vocal fold mobility issues is fibro-optic exams. So the fibro-optic endoscopy is considered to be a quick bedside evaluation tool that is very sensitive and specific for identification of vocal fold mobility issues. And it is a minimally invasive procedure. It helps with direct visualization of vocal cord mobility. And it is limited by excessive movement of the patient, any secretions that can be an obstruction to visualization, and floppy subproglottic structures that can often be associated with a lot of the syndromic patterns that we see with congenital heart disease, including like degeorge and trisomy 21, et cetera. The other major thing that has been described is interoperator reliability for fibro-optic exams. And it has been shown that without any accompanying audio or any of the preceding clinical presentation or symptomatology, the interoperator reliability of fibro-optic endoscopy in identifying vocal fold dysfunction really plummets. So that's what has brought around this new advent of vocal cord ultrasounds as a driver for identification of vocal fold mobility issues. And I want to highlight this recent paper from Pediatric Critical Care Medicine from the Research Collaborative on Critical Care Ultrasound that did a quick systematic review of what data there is available and what they found that it is non-invasive, less likely to perturb hemodynamics and it is about 91% sensitive and 97% specific for identification of vocal cord dysfunction. The interesting thing that they found in this that we wanted to address with our work was that most of these existing studies are the vocal cord ultrasounds and laryngeal ultrasounds are still performed by radiologists or sonographers. And they all utilize like the matrix probes that are available with the radiological machines. And only one particular study from the Middle East that utilized the critical care physicians with point-of-care ultrasound machines for the laryngeal ultrasound screening. So the next question we asked ourselves when going into the background is, can this be a POCUS tool that can be done at the bedside by the critical care physician? It is superficial, it is easy to perform, it uses the standard linear probes, there is a flat learning curve associated with learning how to do the procedure and it is shown to have good inter-rate reliability in the existing literature. I just threw up an example visualization of the larynx from one of our studies and just to orient people who are not familiar with it, just going through the individual structures, the center portion here is the tracheal column, the thyroid cartilage is the one that is more superficial and close to the probe. There are the vocal cords on either sides can be identified here and then there's the arytenoid folds on either side, which is very important for a quantification of vocal cord dysfunction. So like I said before, the assessment can be qualitative or quantitative and for the quantitative portion of things, what we use is the arytenoid fold angle and as illustrated by this patient who had a right-sided vocal cord dysfunction, you can see that the adduction is impaired and therefore the arytenoid angle is definitely larger on the side where there is a vocal cord dysfunction. So with that background, our main objective for this study was to compare the performance of vocal cord ultrasound when compared to fibro-optic endoscopy in post-cardiac surgery patients. The way we did that at Boston Children's was we implemented a clinical algorithm based on our institutional experience and after discussing with our surgical and otolaryngology colleagues and we started by identifying the patients at risk for feeding dysfunction and the screening criteria was procedures that were related to the aortic arch as described before. So any patients who had a major aortic arch procedure or a vascular ring repair procedure, a stage one or stage two operation for hypoplastic left heart syndrome or any complex airway repair were included in our algorithm. Once we screened in the patient, the patient would get an ORL evaluation, ideally within 72 hours after extubation and the patient being off positive pressure ventilation and depending on what the ORL exam showed, if they had a normal exam, they would exit the algorithm. If they had ongoing feeding concerns, they would still get a feeding consult and from there on, assess their risk for aspiration. If they had an abnormal vocal cord exam, it will automatically trigger a feeding team evaluation followed by an evaluation for quantification of aspiration risk with a possible modified barium swallow. If they had an abnormal head and neck exam, prior to their ORL scope, they would get an additional evaluation to assess if there is any impediment to the fibro-optic endoscopy. So in this particular algorithm, the way we fit in the vocal cord ultrasound was to do it concomitantly with the ORL evaluation with the fibro-optic endoscopy. So in terms of the equipment and personnel, we used our standard available FUJIFILM Sonosight machine. We used the standard linear 19.5 megahertz probe. Four of the CICU physicians, three attendings and one fellow completed the online training module that I have linked below. This was basically a YouTube and written module that we completed, that was built by the Texas Children's team. And after training ourselves on the module, we started screening patients per algorithm and then they would get screened in for the fibro-optic endoscope. Based on the clinician availability, they would also get a vocal cord ultrasound by one of the four of the people. Moving into what we saw, in terms of what kind of patients we included in the initial phase of the study, there were about 63 patients at the time of us doing our interim analysis, and of which majority of the patients had a co-octation repair, either through a median sternotomy or a thoracotomic approach. And the second most common was the interrupted aortic arch repair, and then the other one was the vascular ring or airway repair. There was some stage one palliations, stage two palliations and complex aortic repairs. And there was one patient with an aortic aneurysm repair. We compared our algorithm effectiveness independently of the vocal cord ultrasound screening, and we compared it to the initial data that we had pre-implementation of the algorithm. And what we noticed that the percentage of patients who were taking at least some portion of their FEEDS-PO at discharge went up significantly from 22% to 90% after the implementation of the algorithm. The median length of stay in the ICU decreased from 9.5 to seven days, and median hospital length of stay did decrease significantly to 14 days from 41.4 days. Though there might be other confounding factors that may have contributed to it in terms of our total improvement in our practice. Median time to initial screening for an ENT exam came down to six days from 10 days after implementation of the algorithm. And the number of patients who were discharged with a durable form of enteral axis or a nasoenteral feeding tube came down to 23% from 38%. In terms of the vocal cord ultrasounds themselves, the incidence of vocal cord dysfunction in the fibro-optic endoscopy group and vocal cord ultrasound group was similar at 24 patients each. That being said, there was one patient that was identified with fibro-optic endoscopy that we had not diagnosed with vocal cord dysfunction. And there was another patient who had a floppy upper airway that the oral folks were not able to visualize the cords that we were able to identify with vocal cord dysfunction and with vocal cord ultrasounds, which is what makes the number even. There was no incidence of hypoxia defined as drop in the SATs by more than 10 points in the vocal cord ultrasound group compared to two patients in the fibro-optic endoscopy group. There was increase in respiratory rate, which was defined as increase in the respiratory rate by at least 10 points. It happened about seven times in the fibro-optic endoscopy group as opposed to just one patient in the vocal cord ultrasound group. There was no incidence of bradycardia or hypotension. And overall sensitivity was about 98.6% and specificity was about 97.5%. And the average time to screening was definitely lower in the vocal cord ultrasound group when compared to the fibro-optic endoscopy group. And definitely was concomitant with the fact that we had to consult a new service to come evaluate the patient as opposed to the clinicians at the bedside doing it. So moving on to what I think is most interesting to me is the clinical impressions that we learned from this project. It is definitely better tolerated, is technically less challenging, and it has a flat learning curve. And it is not influenced by upper airway tone or patient cooperation. I put an example vital signs trend from our T3 data for a patient who received a fibro-optic endoscope and a vocal cord ultrasound. And the red line indicates the time of the scope or the ultrasound being performed. As you can see, this patient had a tachycardia associated with heart rate raised up to the 190s and was also found to have a desaturation episode immediately following the fibro-optic endoscope. And the time interval displayed here is about 30 minutes, starting with about two minutes preceding the procedure to about 28 minutes after. And you can see for the same patient with the vocal cord ultrasound, there was no tachycardia or hypoxia associated for this patient. And the other thing we noticed is that the inter-rater reliability, the understanding and identification of vocal cord dysfunction is definitely influenced by probe position. Just to give a brief background again, the probe is positioned straight on at the neck at the level of the thyroid cartilage. And this is what we are essentially looking at in terms of the laryngeal ultrasound. The image position and depth affect the interpretation. This is an example patient where you can visualize their right vocal cord very well, but the left vocal cord is not seen as well. And once we center the image and maintain head position, you can identify the left vocal cord much better. And you can see that the adrenoid fold angle is definitely different between the two vocal folds. And this patient definitely has a left vocal cord dysfunction. Probe orientation is also very critical to interpretation. This particular patient was identified as a left vocal cord dysfunction, but in reality, the probe was inverted and the patient was found to have right vocal cord dysfunction in fibroptic exam. So the basics definitely matter. The other thing we noticed is supraglottic anatomy does not play a role in identification with vocal cord ultrasound. This is the fibroptic endoscopy image of the patient with a floppy arytenoid fold. As you can see, you can see the arytenoids very well, the epiglottis very well, but the vocal cords are not really visualized well. The same patient, this is the vocal cord ultrasound, showing significant left-sided dysfunction. Interestingly, on discussing with our oral colleagues, they found that the left side of the arytenoid wasn't moving well, which was concomitant with the side of vocal cord dysfunction. Patient cooperation and technical ease. Older patients can be very cooperative with vocal cord ultrasounds because they can be very much fascinated by the images that we are seeing, and they cooperate in phonation to quantify the dysfunction better, and they're probably less bothered by the fact that there's not a tube sticking down their nose. So this particular patient is a six-year-old with developmental delay who underwent a vocal cord exam with the fibroptic endoscopy first, and then they were trying to get her to enunciate some things so they could identify and quantify the vocal cord dysfunction, but they were not very successful because she was coughing incessantly. And then in the same patient, this is her saying E-O repeatedly, and you can see the bilateral mobility of the vocal cords that are quantified much better. I'm hoping all the images are looking great in there. So to conclude, vocal cord ultrasounds are very specific and sensitive for screening for vocal cord dysfunction. They're well-tolerated by patients and have a low incidence of complications. They can shorten the time to screen for patients with feeding symptoms, and it can be a CIC-implemented POCUS tool. Standard linear probes work well. The learning curve is small based on our experience, and it can be a primary screening tool in many resource-limited settings. In terms of future directions, the question that we are trying to see if we can investigate further is whether a CNN-based image recognition model, which is just another word for machine learning-based AI-driven image recognition model can be helpful for quantification of the vocal cord dysfunction by automatic recognition of the arytenoid fold angles. And can it help in longitudinal assessment for clinical improvement? And can it help with automated training and redirection models for quality improvement of point-of-care ultrasound for vocal cord dysfunction? Just a quick plug to my wonderful team who helped me on this project, and thank you all for listening.

ultrasound screening

vocal cord dysfunction

congenital heart surgery

children

feeding dysfunction