Inter-observer reliability of localization of recorded stridor sounds in children.

Objective: To determine the inter-observer reliability in localization of recorded stridor sounds in children.

Method: The stridor sounds of 28 children programmed for laryngobronchoscopy were recorded with a high quality digital recorder. Nineteen of these recordings with a diagnosis confirmed by endoscopy, were presented to otorhinolaryngology residents, academic specialists and non-academic consultants (n=38) in different situations with and without additional information about the subject. The participants were requested to score the sounds as pharyngeal, supraglottic, glottic, subglottic or tracheal in both situations. The scores were analyzed per group of participants, per location of obstruction and per diagnosis in the different situations.

Results: The performance of the total group was just above chance level with an average score of 29.6%. The total results improved slightly with the additional information present, although not significantly. No significant difference was found between the three categories of participants. The supraglottic sounds were significantly better differentiated from the other locations in both assessment types (p<0.001). The tracheal region (p<0.01) and the supraglottic region (p<0.05) received significantly higher scores when the additional information was present. Laryngomalacia was significantly better differentiated (p<0.001) from the other diagnoses. The improvement in assessment with and without additional information present was significant for laryngomalacia (p=0.002) and tracheomalacia (p<0.035).

Conclusion: The clinical observation of stridor in children shows poor levels of localization. Even though the two most common diagnoses, laryngomalacia and tracheomalacia are localized more correctly, the general performance is not significantly higher than random. No evidence was found in this study for differences in results by the observer's level of experience. We would recommend that an observational policy is only adequate when clear clinical signs are present that indicate laryngomalacia as first differential diagnosis. In other cases rigid and flexible endoscopy is indicated to locate the site and nature of obstruction.