CLINICAL-EVALUATION AND TEST-RETEST RELIABILITY OF THE IHR-MCCORMICK AUTOMATED TOY DISCRIMINATION TEST

The IHR-McCormick Automated Toy Discrimination Test (ATT) measures the minimum sound level at which a child can identify words presented in quiet in the sound field. This 'word-discrimination threshold' provide s a direct measure of the ease with which a child can identify speech and a surrogate measure of auditory sensitivity. This paper describes steps taken to maximize the test-retest reliability of the ATT and to enable it to measure word-discrimination thresholds in noise as well a s in quiet. It then describes the results of a clinical evaluation of the ATT in which paediatric audiologists measured word-discrimination thresholds in quiet from 215 successive attendees (in the age ranee 2 to 13 years) at a paediatric audiology clinic presenting over a 2-mont h period. When children with atypical cognition or delayed development of language were excluded, 72% of the children provided two word-disc rimination thresholds and 83% provided at least one word-discriminatio n threshold. Children who failed to provide word-discrimination thresh olds were generally younger than four years of age. Although a few chi ldren who could not perform pure-tone or warble-tone audiometry manage d to provide word-discrimination thresholds, most children who could p erform the ATT could also perform pure-tone audiometry. The average pu re-tone threshold in the better-hearing ear could be predicted from th e word-discrimination threshold with a 95% confidence interval of +/-1 3 dB. The lest-retest reliability of the ATT was measured in two ways. First, to enable comparison with published results, the within-subjec ts standard deviation of word-discrimination thresholds was calculated . It varied as a function of age and degree of impairment, but was nev er worse than 3.3 dB. Children of four years of age and older displaye d the adult reliability of 2.3 dB. Second, the variability of absolute differences between word-discrimination thresholds was calculated. It was such that a change of 7 dB between two runs-of the lest (e.g. aid ed and unaided) would be expected to occur by chance less than one tim e in 20. These results extend previous evaluations of the ATT to a cli nically representative population and confirm that word-discrimination thresholds provide a useful complement to warble-tone and pure-tone a udiometry.