Professional Documents
Culture Documents
Objective It is hypothesized that laryngeal edema is caused by laryngopharyngeal reflux (LPR) (ie, gastroesoph-
ageal reflux extending into the larynx and pharynx). The validated reflux finding score (RFS) assesses LPR disease in
adults. We, therefore, aimed to develop an adapted RFS for infants (RFS-I) and assess its observer agreement.
Study design Visibility of laryngeal anatomic landmarks was assessed by determining observer agreement. The
RFS-I was developed based on the RFS, the found observer agreement, and expert opinion. An educational tutorial
was developed which was presented to 3 pediatric otorhinolaryngologists, 2 otorhinolaryngologists, and 2 gastro-
enterology fellows. They then scored videos of flexible laryngoscopy procedures of infants who were either diag-
nosed with or specifically without laryngeal edema.
Results In total, 52 infants were included with a median age of 19.5 (0-70) weeks, with 12 and 40 infants, respec-
tively, for the assessment of the laryngeal anatomic landmarks and the assessment of the RFS-I. Overall interob-
server agreement of the RFS-I was moderate (intraclass correlation coefficient = 0.45). Intraobserver agreement
ranged from moderate to excellent agreement (intraclass correlation coefficient = 0.50-0.87).
Conclusion A standardized scoring instrument was developed for the diagnosis of LPR disease using flexible
laryngoscopy. Using this tool, only moderate interobserver agreement was reached with a highly variable intraob-
server agreement. Because a valid scoring system for flexible laryngoscopy is lacking up until now, the RFS-I and
flexible laryngoscopy should not be used solely to clinically assess LPR related findings of the larynx, nor to guide
treatment. (J Pediatr 2014;165:479-84).
L
aryngomalacia is the most common congenital anomaly of the larynx accounting for over 60% of all noninfectious stri-
dors in infancy.1 In infants with laryngomalacia, laryngeal edema is frequently seen during laryngoscopic examination. It
is commonly thought to be the result of gastroesophageal reflux (GER) extending into the larynx and pharynx: laryng-
opharyngeal reflux (LPR).2-6 However, evidence for this causality is lacking. It is hypothesized that a partial obstruction of the
airway because of laryngomalacia causes a negative intrathoracic pressure, facilitating LPR to occur, which in turn causes laryn-
geal edema. This subsequently leads to a vicious circle of increased obstruction, more LPR, and edema.7
In adults, LPR can reliably be detected using flexible laryngoscopy and a validated reflux finding score (RFS).8 In children,
evaluation of the larynx with flexible laryngoscopy might be hampered by the smaller anatomic landmarks and smaller endo-
scopes with lower image resolution. Furthermore, reflux patterns in infants differ significantly from those in adults as a result of
different feeding patterns and posture. In infants, common endoscopic findings suggested to be LPR-related are edema and
erythema of the arytenoids, postglottic and vocal fold edema, and erythema.9 Despite the abovementioned differences between
adults and infants, no specific scoring instrument exists for this age group.
Based on medical history, physical examination, and laryngoscopic findings, proton pump inhibitor (PPI) therapy is
commonly initiated when LPR-related laryngeal edema is suspected to be the cause of symptoms.10,11 Contrary to adults,
however, PPI therapy has been proven ineffective for classical GER symptoms in infants such as regurgitation and excessive
crying and are, thus, not approved by the Federal Drug Administration.12 Furthermore, no evidence is available for their
use in LPR-related symptoms in infants.
The aim of this study was to develop a scoring instrument to evaluate signs of LPR seen on flexible laryngoscopy in infants,
the RFS for infants (RFS-I) and to assess its inter- and intraobserver agreement.
479
THE JOURNAL OF PEDIATRICS www.jpeds.com Vol. 165, No. 3
Statistical Analyses
Data were analyzed using IBM SPSS Statistics 19 (SPSS Inc,
Armonk, New York). For the analysis of the development
of the RFS-I, inter- and intraobserver agreement was deter-
mined for all selected landmarks using Fleiss kappa (kappa
further annotated as k). For the RFS-I, inter- and intraob-
server agreement was determined per scoring item of the
RFS-I and for all scoring items combined. For categorical
data, inter- and intraobserver agreement was calculated using
Cohen k (2 observers) and Fleiss k (>2 observers). For ordinal
data, weighted k and the intraclass correlation coefficient
(ICC) were used. Fleiss k was calculated by using a pre-
made syntax for SPSS (available from corresponding author).
Figure 1. Anatomic landmarks (1 = epiglottis, 2 = aryepiglot-
For the assessment of the inter- and intraobserver agreement
tic fold, 3 = cuneiform cartilages, 4 = vocal cords, 5 = aryte- of the RFS-I, we applied the arbitrary but common scale for k
noids, 6 = postcricoid region, 7 = piriform sinus). and ICC values: 0.00 = no agreement, 0.01-0.20 = slight
agreement, 0.21-0.40 = fair agreement, 0.41-0.60 = moderate
480 van der Pol et al
September 2014 ORIGINAL ARTICLES
agreement, 0.61-0.80 = substantial agreement, 0.81- Table I presents all items of the RFS-I, resulting in a total
0.99 = excellent agreement, and 1.00 = perfect agreement. RFS-I score ranging from 0-8.
A k between 0.6 and 1.0 was considered to be valid for use
in research settings and/or clinical practice. Patients
Of the 52 included infants (median age: 19.5 [0-70] weeks),
Results indications for flexible laryngoscopy were stridor (75%),
obstructive sleep apnea syndrome, apparent life threatening
Development of the RFS-I events, and swallowing problems. Age and sex were equally
The process of developing the RFS-I is shown in Figures 1 distributed in the 2 groups. Inter- and intraobserver agree-
and 2. The first step in the development of the RFS-I was ment was not influenced by previous diagnosis of LPR on
the assessment of the anatomic landmarks, which showed the initial clinical flexible laryngoscopy.
good visibility of the aryepiglottic fold and vocal cords.
The second step in the process of the development of the Interobserver Agreement of RFS-I
RFS-I consisted of the selection of items based on the All interobserver agreement data are shown in Table II. The
items included in the original RFS, the visibility scores as overall interobserver agreement of the total RFS-I score was
reported in Figure 2, and expert opinion. Vocal fold moderate (ICC = 0.45). Highest agreement was reached for
edema is one of the adult RFS items. Based on expert the visibility of the vocal cord (k = 0.44) and fair agreement
opinion, this was translated to visibility of the vocal cords was reached for the presence of erythema or laryngeal edema
and was included in the RFS-I as such. Although the (k = 0.28) for all observers. The observers in group 3
aryepiglottic fold was scored to have good visibility, it was showed highest interobserver agreement (ICC = 0.63).
not included in the RFS-I because it is not used in the
original RFS and was not considered discriminative for Intraobserver Agreement of RFS-I
laryngeal edema by expert opinion. Diffuse laryngeal All intraobserver agreement data are shown in Table II.
edema and endolaryngeal mucus, both items on the Overall intraobserver agreement of the total score of the
original RFS, were selected by expert opinion and RFS-I ranged from moderate to excellent (ICC = 0.42-0.87)
included in the RFS-I. Other items on the adult RFS were between the 3 groups. Agreement per scoring item showed
discussed, yet not included in the RFS-I because these a wide variation between observers (k = 0.13-1.00).
items would be too difficult to assess and/or they are Intraobserver agreement on all scorings items was
absent in infants with and without laryngeal edema. consistently highest in group 2.
Figure 2. Flowchart of development of the RFS-I. The left panel depicts the scores of the observer agreement of the anatomic
landmarks of the larynx. In the right panel the items of the RFS are presented. Based on the visibility of the anatomic landmarks of
the infant larynx and the adult RFS items for the RFS-I were selected through expert opinion.
Development of the Reflux Finding Score for Infants and Its Observer Agreement 481
THE JOURNAL OF PEDIATRICS www.jpeds.com Vol. 165, No. 3
Table II. Inter- and intraobserver agreement per item, per group
Interobserver agreement Intraobserver agreement
Total Group 1 Group 2 Group 3 Group 1 Group 2 Group 3
Erythema/hyperemia/laryngeal edema (ICC/k) 0.28 0.21 0.30 0.39 0.70 0.57 0.36
0.50 0.66 0.14
0.33
Vocal cord visibility (k) 0.44 0.37 0.48 0.73 0.57 0.68 0.79
0.50 0.66 0.41
0.28
Endolaryngeal mucus (k) 0.26 0.35 0.80 0.24 0.90 1.00 0.54
0.13 0.75 0.28
0.60
Total score (ICC) 0.45 0.51 0.51 0.63 0.87 0.84 0.73
0.51 0.80 0.50
0.42
In this study, we found moderate interobserver variability 2. Carr MM, Nguyen A, Poje C, Pizzuto M, Nagy M, Brodsky L. Correla-
of the RFS-I, which is likely to be the result of 1 or more of tion of findings on direct laryngoscopy and bronchoscopy with presence
of extraesophageal reflux disease. Laryngoscope 2000;110:1560-2.
several factors. First, it is inherent to the infant larynx that de-
3. El-Serag HB, Gilger M, Kuebeler M, Rabeneck L. Extraesophageal asso-
tails of the larynx are less visible compared with adults, purely ciations of gastroesophageal reflux disease in children without neuro-
because of size of the anatomic landmarks. Directly related is logic defects. Gastroenterology 2001;121:1294-9.
the fact that smaller endoscopes produce images with less 4. Matthews BL, Little JP, Mcguirt WF, Koufman JA. Reflux in infants with
optimal resolution. Second, observers in this study annotated laryngomalacia: results of 24-hour double-probe pH monitoring. Oto-
laryngol Head Neck Surg 1999;120:860-4.
that the limited visibility of details on the examination videos
5. Halstead L. Gastroesophageal reflux: a critical factor in pediatric subglot-
may be inherent to infants being awake and agitated. Flexible tic stenosis. Otolaryngol Head Neck Surg 1999;120:683-8.
laryngoscopy, at least in our hands, is performed in an awake 6. Bibi H. The prevalence of gastroesophageal reflux in children with tra-
patient in order to study the larynx dynamically. Third, the cheomalacia and laryngomalacia. Chest 2001;119:409-13.
observers in our study mentioned having difficulties inter- 7. Landry AM, Thompson DM. Laryngomalacia: disease presentation,
spectrum, and management. Int J Pediatr 2012;2012:753526.
preting endolaryngeal mucus. This item, which was derived
8. Belafsky PC, Postma GN, Koufman JA. The validity and reliability of the
from the adult RFS, could have influenced the moderate reflux finding score (RFS). Laryngoscope 2001;111:1313-7.
outcome of the overall observer variability. 9. May JG, Shah P, Lemonnier L, Bhatti G, Koscica J, Coticchia JM.
For the assessment of the items included in the RFS-I, it is Systematic review of endoscopic airway findings in children with
conceivable that images from anesthetized infants may suffice. gastroesophageal reflux disease. Ann Otol Rhinol Laryngol 2011;
120:116-22.
Although it is frequently recommended to perform flexible
10. Hicks DM, Ours TM, Abelson TI, Vaezi MF, Richter JE. The prevalence
laryngoscopy in infants,19 it could be hypothesized that exam- of hypopharynx findings associated with gastroesophageal reflux in
ination of the infant larynx with direct laryngoscopy during normal volunteers. J Voice 2002;16:564-79.
general anesthesia may provide a better agreement of the 11. Barron JJ, Tan H, Spalding J, Bakst AW, Singer J. Proton pump in-
RFS-I. However, the gain of better perception of details, hibitor utilization patterns in infants. J Pediatr Gastroenterol Nutr
2007;45:421-7.
must outweigh the impact for the need for general anesthesia.
12. Chen IL, Gao WY, Johnson AP, Niak A, Troiani J, Korvick J, et al. Proton
To test the association between LPR disease and laryngeal pump inhibitor use in infants: FDA reviewer experience. J Pediatr Gas-
aberrations one would ideally be able to perform a random- troenterol Nutr 2012;54:8-14.
ized controlled double blinded intervention trial in infants 13. Hartl TT, Chadha NK. A Systematic Review of Laryngomalacia and acid
with LPR using objective and reproducible means of detect- reflux. Otolaryngol Head Neck Surg 2012;147:619-26.
14. Thompson DM. Abnormal sensorimotor integrative function of the lar-
ing LPR. The sensitivity and specificity of this method can
ynx in congenital laryngomalacia: a new theory of etiology. Laryngo-
then be calculated in hindsight based on outcome data. scope 2007;117:1-33.
We have now shown that flexible laryngoscopy using the 15. Van der Pol RJ, Smits MJ, van Wijk MP, Omari TI, Tabbers MM,
RFS-I is not suitable for this purpose. Several other studies Benninga MA. Efficacy of proton-pump inhibitors in children with
have been performed to come up with normative data for gastroesophageal reflux disease: a systematic review. Pediatrics 2011;
127:925-35.
pharyngeal and proximal esophageal pH and pH-
16. De Bortoli N, Nacci A, Savarino E, Martinucci I, Bellini M, Fattori B,
impedance values in adults.20-24 However, debate still re- et al. How many cases of laryngopharyngeal reflux suspected by laryn-
mains on pathologic pH cut-off values, position of the pH goscopy are gastroesophageal reflux disease-related? World J Gastroen-
sensor and interobserver variability of pharyngeal terol 2012;18:4363-70.
impedance measurements.24-26 In infants, no validated 17. Belafsky PC, Postma GN, Koufman JA. Validity and reliability of the re-
flux symptom index (RSI). J Voice 2002;16:274-7.
questionnaires or normal values of pharyngeal pH or pH-
18. Eherer AJ, Habermann W, Hammer HF, Kiesler K, Friedrich G, Krejs GJ.
impedance are available for detecting LPR disease. Data in Effect of pantoprazole on the course of reflux-associated laryngitis: a
infants are unlikely to resemble adult data because physio- placebo-controlled double-blind crossover study. Scand J Gastroenterol
logic LPR is more likely to occur as frequent regurgitation 2003;38:462-7.
without any other symptoms or complications is present 19. Yellon RF, Borland LM, Kay DJ. Flexible fiberoptic laryngoscopy in chil-
dren: effect of sitting versus supine position. Int J Pediatr Otorhinolar-
in the majority of infants.27 Therefore, the first step ahead
yngol 2007;71:1293-7.
should be to perform studies aiming to find a objective 20. Sun G, Muddana S, Slaughter JC, Casey S, Hill E, Farrokhi F, et al. A new
and reproducible diagnostic approach for infants suspected pH catheter for laryngopharyngeal reflux: normal values. Laryngoscope
to have LPR disease. n 2009;119:1639-43.
21. Dobhan R, Castell DO. Normal and abnormal proximal esophageal acid
exposure: results of ambulatory dual-probe pH monitoring. Am J Gas-
Submitted for publication Oct 22, 2013; last revision received Apr 21, 2014;
accepted May 12, 2014. troenterol 1993;88:25-9.
22. Bove M, Ruth M, Cange L, M ansson I. 24-H pharyngeal pH monitoring
Reprint requests: Rachel J. van der Pol, MD, Department of Pediatric
in healthy volunteers: a normative study. Scand J Gastroenterol 2000;35:
Gastroenterology and Nutrition, Emma Children’s Hospital/Academic Medical
Center, Meibergdreef 9, C2-312, 1105 AZ Amsterdam, The Netherlands. 234-41.
E-mail: r.j.vanderpol@amc.nl 23. Andersson O, Ylitalo R, Finizia C, Bove M, Magnus R. Pharyngeal reflux
episodes at pH 5 in healthy volunteers. Scand J Gastroenterol 2006;41:
138-43.
References 24. Zerbib F, Roman S, Bruley Des Varannes S, Gourcerol G, Coffin B,
Ropert A, et al. Normal values of pharyngeal and esophageal 24-hour
1. Edmondson NE, Bent JP, Chan C. Laryngomalacia: the role of gender pH impedance in individuals on and off therapy and interobserver
and ethnicity. Int J Pediatr Otorhinolaryngol 2011;75:1562-4. reproducibility. Clin Gastroenterol Hepatol 2013;11:366-72.
Development of the Reflux Finding Score for Infants and Its Observer Agreement 483
THE JOURNAL OF PEDIATRICS www.jpeds.com Vol. 165, No. 3
25. Habesoglu M, Habesoglu TE, Gunes P, Kinis V, Toros SZ, Eriman M, et al. of the role of acid and pepsin in the development of laryngeal. Laryngo-
How does reflux affect laryngeal tissue quality? An experimental and histo- scope 1991;101:1-78.
pathologic animal study. Otolaryngol Head Neck Surg 2010;143:760-4. 27. Campanozzi A, Boccia G, Pensabene L, Panetta F, Marseglia A,
26. Koufman JA. The otolaryngologic manifestations of gastroesophageal Strisciuglio P, et al. Prevalence and natural history of gastroesoph-
reflux disease (GERD): a clinical investigation of 225 patients using ageal reflux: pediatric prospective survey. Pediatrics 2009;123:779-
ambulatory 24-hour pH monitoring and an experimental investigation 83.
S ome clinical “facts” are uncontested; for example that pulmonary artery pressure and resistance fall in the first few
days after birth. Nonetheless, in the early 1960s, this was suspected but not firmly known and Emmanouilides et al
figured it out. How? By performing umbilical artery catheterizations on sleeping healthy newborns with 5F nasogastric
tubes, passing them to the thoracic descending aorta and into the pulmonary artery via the ductus arteriosus. Pressure
recordings and oximetry were used to define catheter location and provide study measurements. Elegant graphs
showed the decline of pulmonary artery pressure and resistance day by day as well as the intrinsic variability of the
rate of decline. Parental consent was obtained, but in this era institutional review boards did not uniformly exist.
As arresting as the narrative of this paper are the references, which highlight contemporary investigations in newborns
and in the laboratory that tease out elements of the transitional circulation and the physiology behind the new diag-
nosis of persistent pulmonary hypertension of the newborn. It is perhaps no accident that the authors of this paper,
and the key references, produced the first truly authoritative textbooks in the field of pediatric cardiology.
The contrast between the investigative era of this study and today are striking. One wonders if institutional review
board approval could be obtained; morbidity to the study subjects is not reported. Nonetheless, families trusted the
study physicians and one assumes that the families believed the knowledge obtained from the study would benefit
many children other than their own, which it has. The skill of the investigators was high, performing bedside cathe-
terizations without waking sleeping infants and without fluoroscopic guidance, recording physiologic data, and ob-
taining oximetry data with equipment far inferior and slower than contemporary devices.
The lesson I take from rereading this seminal paper is to leave no piece of knowledge in medicine unexamined. The
knowledge came from somewhere: pioneering efforts that might now be considered unethical, medical mistakes that
under scrutiny led to revisions in technique, and, perhaps most important, imaginative thinking about how to use
existing tools to answer clinical questions deemed critical. Today, pulse oximetry and echocardiography can provide
similar (though less accurate) data; nonetheless, we interpret these studies based on “what we already know.” Well, 50
years ago, we did not know for sure the time course of change in pulmonary artery pressure.
Samuel S. Gidding, MD
Nemours Cardiac Center
A. I. DuPont Hospital for Children
Wilmington, Delaware
http://dx.doi.org/10.1016/j.jpeds.2014.02.058