Effects of early test termination in a German matrix speech test in noise in cochlear implant recipients.

Abstract

Background: Speech reception thresholds (SRTs) can be estimated with speech in noise tests using adaptive procedures. Since these tests are time-consuming, it is preferable to use as few trials as possible to identify reliable results.

Aims/objectives: To estimate the impact of early test termination in a German matrix speech test in noise (Oldenburger Satztest) on SRT estimation after 20 instead of after 30 sentences and to identify groups of subjects, for who such a tradeoff between measurement time and accuracy might be useful and acceptable.

Material and methods: Retrospective analysis of 789 tests from cochlear implant users and from simulated tests in 10000 virtual subjects.

Results: Early test termination results in increased estimation errors of the SRT in both real, and simulated listeners. Average test time is reduced from 5.0 to 3.6 min. For estimated SRT's above approximately +2 dB after 20 test sentences, the expected average difference to the result after 30 trials becomes higher than 0.5 dB.

Conclusions and significance: If adaptive speech in noise tests are terminated already after 20 sentences, time can be saved, but it should be considered only in listeners with relatively favourable SRT-estimates after 20 trials, preferably below +2 dB.