Abstract
The aim of the study was to determine the test-retest reliability of MMN and LDN recorded to simple speech contrasts in children with listening difficulties. MMN and LDN responses were recorded from Fz and Cz electrodes for a /da/-/ga/ contrast twice within a 10-day period. To extract MMN and LDN, auditory-evoked responses to /ga/ stimuli presented alone were subtracted from the responses to /ga/ presented within an oddball sequence. Intraclass correlation coefficients (ICCs) were used to determine test-retest reliability of MMN and LDN. Eighty-five children aged 7.0-12.8 years were grouped into four clusters; Cluster 1 included children with global difficulties (n = 34); Cluster 2 had children with poor auditory processing but relatively good word reading (n = 19); Cluster 3 had poor auditory processing, memory and attention skills but relatively good nonverbal intelligence and language skills (n = 14); and Cluster 4 had poor auditory processing and attention but relatively good memory skills (n = 18). At Visit 1, MMN and LDN were detectable at Cz in only 42% and 21% of participants, respectively. The ICC for the four clusters for MMN ranged from 0.36 to 0.76; LDN ICCs were 0.21-0.54. MMN was most replicable (71%) for children with listening difficulties in Cluster 3 with good nonverbal intelligence and language. The results do not support the clinical utility of MMN or LDN for objective assessment of auditory discrimination. Although MMN had better test-retest reliability, overall detectability was poor. Better detectability is required for any clinical utility in children.