Development of an Algorithm for the Diagnosis of Otitis Media

Background lative importance of signs and symptoms in the diagnosis of otitis media has not been adequately evaluated. This has led to a large degree of variation in the criteria used to diagnose otitis media, which has resulted in inconsistencies in clinical care and discrepant research findings. s p of experienced otoscopists examined children presenting for primary care. We investigated the signs and symptoms that these otoscopists used to distinguish acute otitis media (AOM), otitis media with effusion (OME), and no effusion. We used recursive partitioning to develop a diagnostic algorithm. To assess the algorithm, we validated it in an independent dataset. s g of the tympanic membrane (TM) was the main finding that otoscopists used to discriminate AOM from OME; information regarding the presence or absence of other signs and symptoms added little to the diagnostic process. Overall, 92% of children with AOM had a bulging TM compared with 0% of children with OME. Opacification and/or an air-fluid level was the main finding that the otoscopists used to discriminate OME from no effusion; 97% of children diagnosed with OME had an opaque TM compared with 5% of children diagnosed with no effusion. An algorithm that used bulging and opacification of the TM correctly classified 99% of ears in an independent dataset. sions g of the TM was the finding that best discriminated AOM from OME. The algorithm developed here may prove to be useful in clinical care, research, and education concerning otitis media.