Limiting asymmetric hearing improves benefits of bilateral hearing in children using cochlear implants

Neurodevelopmental changes occur with asymmetric hearing loss, limiting binaural/spatial hearing and putting children at risk for social and educational challenges. These deficits may be mitigated by providing bilateral hearing in children through auditory prostheses. Effects on speech perception and spatial hearing were measured in a large cohort of >450 children who were deaf and used bilateral cochlear implants or bimodal devices (one cochlear implant and a contralateral hearing aid). Results revealed an advantage of bilateral over unilateral device use but this advantage decreased as hearing in the two ears became increasingly asymmetric. Delayed implantation of an ear with severe to profound deafness allowed asymmetric hearing, creating aural preference for the better hearing ear. These findings indicate that bilateral input with the most appropriate device for each ear should be provided early and without delay during development.


Longitudinal changes in asymmetry in speech perception
Supplemental Figure 1. Longitudinal changes in asymmetry in speech perception with age. The difference in speech perception (asymmetry) in quiet between the first implanted ear and second implanted or hearing aid ear did not change with age at test. Coloured lines join repeated results for each child; dashed black lines indicate non-significant (p<0.05) group changes in asymmetry with age based on linear mixed-effects regression. The number of children (n) is provided for each group.

Bilateral advantage of speech perception
Bilateral advantage over listening with one ear in noise is shown in Supplemental Figure   2a, and over listening with CI2/HA alone in both quiet and noise is shown in Supplemental Figure 2b. Bilateral advantage over listening with only the HA/CI2 was greater than over listening with CI1 alone (F(1,87)=7.8, p=0.007) and there was no interaction with group (F(4,87)=2.1, p=0.09), reflecting overall asymmetry between unilateral conditions. There was a main effect of group (F(4,87)=3.5, p=0.011), whereby average bilateral benefit for simultaneous users was 12.2±4.2 RAU less than sequential users (z=-2.9, p=0.026) and 12.6±3.9 RAU less than bimodal users (z=-3.2, p=0.011). The bilateral advantage over listening with HA/CI2 was not different in quiet versus noise (F(1,95)=0.0, p=0.82), but there remained a significant effect of group (F(4,95)=3.4, p=0.011). Average bilateral advantage over CI2 (left ear) alone for simultaneous users was 13.5±4.6 RAU less than sequential users (z=-3.0, p=0.025) and 14.1±4.4 RAU less than the bilateral advantage over the HA alone of bimodal users (z=-3.2, p=0.012).

Supplemental Figure 2. Bilateral advantage for speech perception in noise. (a)
The difference in speech perception in rationalized arcsine units (RAU) while testing in noise for the bilateral condition over listening with each ear alone. Bilateral advantage over listening with only the HA/CI2 was greater than over listening with CI1 alone. Average bilateral benefit for simultaneous users was less than sequential and bimodal users. (b) Bilateral advantage over listening with HA/CI2 alone in both quiet and noise. Bilateral advantage differed by group but not by whether there was noise present. Average bilateral advantage over CI2 alone for simultaneous users was less than sequential users and less than the bilateral advantage over the HA alone of bimodal users.
Supplemental Figure 3 shows the advantage of bilateral input to speech perception over listening with only CI1. As the asymmetry between unilateral scores increasingly favoured CI1 (positive asymmetry values), the advantage of adding HA/CI2 decreased (all FDR-corrected p<0.05). For sequential and some older simultaneous users who had large asymmetries favouring CI1, adding CI2 decreased speech perception accuracy. On the other hand, for bimodal and some simultaneous users who had asymmetries favouring the HA/CI2 (left CI for simultaneous group), the bilateral advantage was greatest because the better perceiving ear was given access to sound. This negative correlation between asymmetry and bilateral advantage also occurs when considering bilateral advantage over the best performing ear: advantage to providing bilateral input decreased as absolute asymmetry increased (p<0.05 for all groups; Supplemental Figure 3). Figure 3. Best advantage to bilateral input with symmetric hearing. The advantage to speech perception in quiet was calculated as the difference between scores with bilateral input and the first implanted ear. This bilateral advantage decreased as the absolute value of the asymmetry in speech perception between ears increased. Asterisks indicate significant correlation after FDR corrections for multiple comparisons (p<0.05).

Test delivery affects speech perception scores in noise but not measures of asymmetry or bilateral advantage
To determine the effect of test delivery on speech perception measures, repeated measures ANOVA was completed on a subset of children (181/288 = 62.8%) who were tested in both quiet and noise with the most common and challenging test (PBK), using ear (bilateral, CI1, CI2/HA) or difference (asymmetry, bilateral advantage over CI1, bilateral advantage over CI2/HA) as a within-subject factor, method (monitored-live voice, recorded) and group as between-subject factors, and age as a covariate. Recorded speech stimuli were used in 123/181 (68.0%) children tested in quiet and 100/141 (70.9%) children tested in noise. Speech-weighted noise was used with monitored-live voice delivery and when multi-talker noise could not be used with recorded stimuli due to technical difficulties. However, most of the time multitalker noise was used with recorded PBK delivery. The number of children by group, condition, ear(s) tested and delivery method are provided in Supplemental  Figure   6. In quiet, method of stimulus delivery did not significantly affect speech perception scores