Long-term repeatability of the pure-tone hearing threshold and its relation to noise exposure.

Audiometric repeatability is examined for a population of 356 male and female industrial employees (712 ears), sub-divided according to the amount of previous noise exposure. Re-tests were carried out after an interval averaging 13 months. After a similar interval, 150 of the subjects gave a third audiogram. Fixed-frequency self-recording audiometry was employed, using common equipment and procedures throughout. Results are presented as distributions of the algebraic test-re-test differences for each frequency and for the frequency combination 1-2-3 kHz. Repeatability was best at 1 and 2 kHz and substantially poorer at 6 kHz. Cumulative distributions of the signless differences show that more than 50% of initial hearing threshold levels repeated to better than 5 dB. For all frequencies except 6 kHz the mean values of signless test-re-test differences for the various sub-groups were between 3.4 and 5.6 dB; at 6 kHz the values were between 5.8 and 7.8 dB. In the case of the three-frequency average, the mean differences were reduced to between 2.8 and 3.8 dB. A few repeats (about one in 200) stand out as clearly anomalous; these occur only in isolation and mainly at the higher frequencies. The performance of groups classified by the amount of their previous noise exposure did not differ significantly. In the case of those tested three times, the repeatability over the second inter-test interval showed at most a marginal improvement compared with the first interval.