A recent study by West & Harris (1983) examined the duration of episodes of otitis media with effusion (OME) in 7 and 8 year old children. It was found that 58% of episodes lasted less than 5 weeks. Among the 88 children there were 125 separate episodes of OME.
The results as presented are not in a suitable form to answer a currently germane question: the best retest criterion to choose for a tympanometric screening programme.
It has been recognised for some time that impedance audiometry is a sensitive detector of OME, but since OME is a labile disease, single testing using tympanometry will result in over-referral of children suffering from a short-duration isolated episode. Retest intervals from 4 weeks to 3 months have been recommended in the literature. The raw data from West & Harris' study were examined to try and provide an empirical base for choice of a suitable retest interval, to evaluate the accuracy of a single test compared with two tests as a predictor of subsequent test results, and to provide guidance in the use of results from one or both ears in determining whether a given test was passed or failed.
The data were examined and a pass or fail on either one or both ears was recorded for each of up to 17 test occasions at 2-weekly intervals over the school year, for each of the 100 subjects. The ratios of test occasions passed using one ear or two ears criteria were evaluated for each subject. That is, an index was developed describing the frequency with which each child passed tympanometry.
The ideal screening test would clearly differentiate between those children who are chronic sufferers of OME from those who are predominantly healthy but who occasionally suffer short bouts of OME. As can be seen from Figure 1, there is a small group of children who are always or nearly always disease-free (approximately 40% using a 1-ear criterion, and 70% using 2-ears). There is a small core of children with continuous OME (23% using a 1-ear criterion, and 10% using 2-ears), with the rest fluctuating throughout the year.
Figure 1. Number of children having different OME-free Indices, as a function of 1- or 2-ear criteria.
The subjects were divided into three groups depending on the outcome of the first test: those who passed both ears, those who failed 1 ear, and those who failed both ears. The indices for the total year's results were examined for each group (see Figures 2 and 3). Of the children who initially passed both ears, the vast majority passed all or nearly all subsequent tests. A similar pattern emerged for those who failed only 1 ear on the first test, when subsequent results using a 2-ear failure criterion are examined. That is, subjects who fail the first test on 1 ear only are only slightly more likely to suffer binaural OME subsequently. Of subjects who failed both ears on the first test, half had indices less than 0.2. Over 70% failed all test occasions on at least 1 ear.
Figure 2. Percent of children with various OME-free Indices (based on a 2-ear failure criterion) who initially passed, or failed on 1 or 2 ears.
Figure 3. Percent of children with various OME-free Indices (based on a 1-ear failure criterion) who initially passed, or failed on 1 or 2 ears.
These results demonstrate that if a failure criterion of 2 ears is used, a single test result can be a fairly powerful predictor of subsequent results. Agreement within this sample is higher than has been reported elsewhere, probably because the subjects were of an age where OME episodes are reduced in frequency.
If a retest interval of 8 weeks is used, the outcome (based on a failure criterion of 2 ears) is as shown in Table I.
Retesting children who initially fail would therefore reduce the number to be referred for medical intervention by four (22%).
Examination of the overall indices of the subjects in each cell shows that although at face value there is a high rate of misses, these subjects are not chronic OME sufferers.