A model to refine the selection of children with otitis media with effusion for adenoidectomy

The selection of children with otitis media with effusion for adenoidectomy has been criticized on a number of grounds. In spite of the widespread popularity of the procedure there are few objective guidelines to help the otolaryngologist make a decision which is therefore usually subjective. We present a scientifically derived model using a Generalized Linear Interactive Modelling (GLIM) technique from a population of 122 children who underwent either adenoidectomy or no pharyngeal surgery. The outcome parameter studied was otoscopic clearance in an unoperated ear 1 year after treatment. This was related principally to the operation type (P < 0.001) and to the age at which the operation was performed (P < 0.005). Using a probability of clearance threshold of 50% the model was tested on a prospective population of 166 children and outcome was correctly predicted in over 72%. This model enables the surgeon to determine the probability of clearance of effusion following adenoidectomy in an individual child. At a probability threshold of 50% adenoidectomy would be offered to children with bilateral otitis media with effusion aged between 4.33 and 8.00 years. Over the age of 8 years the probability of clearance occurring spontaneously increases beyond 50%. By careful selection of the probability of clearance thresholds it was possible to produce otoscopic resolution in over 80% of those selected to undergo adenoidectomy. The model also enables the otolaryngologist to determine the proportion of children with established disease that require adenoidectomy at his own chosen probability thresholds which will ensure that a more effective resource allocation is achieved.