P19Session 1 (Thursday 12 January 2023, 15:30-17:30)Early and late progress in speech recognition by cochlear implant users
Cochlear implant (CI) outcomes are known to be variable and difficult to predict. According to results from vocoder-based CI simulations with normal listeners, 10 to 20 channels allow >90% sentence recognition in quiet with only tens of minutes of exposure. A few CI recipients may also achieve this high level of speech understanding within a day, and some more still within one month (James et al., Ear Hear 2019, doi:10.1097/AUD.0000000000000670). For others, it appears that several months of experience is required, or they may never reach this level. We hypothesized that CI subjects with no ‘bottom-up’ or ‘top-down’ limiting factors would achieve >90% sentence understanding at one month after activation. We also wanted to understand the progression of sentence understanding in quiet and in noise over time, and how this is influenced by limiting factors.
Longitudinal sentence recognition scores were collected for 32 adult Nucleus CI recipients. Top-down tests were the Montreal Cognitive Assessment (MoCA), the Stroop test for interference, and two subtests from the French ECLA-16+ reading battery to evaluate phonological awareness. Bottom-up tests were based on electrically evoked action potentials (ECAP) performed on a mid-basal and a mid-apical electrode contact: amplitude growth functions (AGF) for anodic- and cathodic-leading biphasic pulses, spread of excitation (SOE) and recovery functions. Sound processor usage was also collected.
CI subjects ranged in age from 26-87 years (median 70). Only four subjects used their processor less than 10 hours per day. A third (13/32) of subjects scored >90% sentence recognition at one month. Six subjects had below normal (<22/30) MoCA scores, five registered interference, and six poor phonological awareness. Subjects with no limiting top-down factors and no limiting bottom-up factors covering both apical and basal electrodes were significantly more likely to achieve >90% score in quiet (Fisher test, odds ratio 10.9, p<0.005). The relationship between limiting factors and scores at one day after activation was not as strong as that seen for one month. However, over several months to a year those with limiting factors scored lower in noise, even where >90% scores were achieved in quiet, compared with those with no identified limiting factors.
These results suggest that top-down factors related to cognition strongly influence speech recognition performance in quiet and in noise with CIs. The presence of bottom-up limiting factors related to neural health or the electrode-neural interface was rare in our sample, or our diagnostic tests were not able to detect them.