Sensitivity, specificity and predictive value of tympanometry in predicting a hearing impairment in otitis media with effusion

Children with otitis media with effusion (OME) need monitoring over time to identify persistence of disease and to assess their bearing thresholds as a surrogate of auditory disability. It would be useful if tympanometry could be used to predict those with an impairment. This study looked at 1153 children, aged between 3.25 and 6.75 years, referred for suspected OME. The inclusion of type C-2 tympanograms with type B tympanograms versus all other tympanogram types increased the sensitivity of detecting air-conduction thresholds and air-bone gaps of various magnitudes in the better hearing ear but lowered the specificity and the positive predictive value. Extending the pressure range to -600 daPa and excluding the 30% of children with poor concentration on audiometry made no difference to the sensitivity and specificity. The results were the same for the poorer hearing ear. These findings are of practical help in monitoring children with OME. Thus taking an air-conduction average of 25 dB HL in the better ear as the level needing detection, if all children are audiometrically assessed then 100% of those with an impairment will be identified. Limiting audiometry to those with a bilateral type B tympanogram reduces the workload by 50%, but 90% of impaired children will still be detected. Limiting audiometry to those with type B or C-2 tympanograms reduces the workload to 69% of the sample, and 95% of impaired children will be identified. With such data, decisions as to how to allocate limited audiometric resources for monitoring children with OME are made easier.