The effect of side of implantation on unilateral cochlear implant performance in patients with prelingual and postlingual sensorineural hearing loss: A systematic review

Cerebral lateralisation of language processing leads to a right ear advantage in normal hearing subjects. The aim of this study was to present a systematic overview of the effect of implantation side on postoperative cochlear implant performance in patients with symmetrical severe to profound sensorineural hearing loss.

The first assumption is that one hemisphere in the brain is of greater influence on brain functions such as speech and language processing compared to the contralateral hemisphere. 4 For speech perception and speech production, left hemisphere dominance is seen in 95-98% of right-handed and in 70-80% of left-handed normal hearing (NH) subjects. [5][6][7] In contrast, for prosodic language functions such as intonation and accentuation, right hemisphere dominance is seen in the majority of these subjects. Overall, the majority of people have a dominant left hemisphere for speech and language processing.
The second assumption is that although the auditory cortex receives auditory input from both ears, it is most strongly stimulated by the contralateral ear. [8][9][10][11][12] Several functional magnetic resonance imaging studies [11][12][13] have revealed that auditory input in one ear predominantly projects to the contralateral superior olivary complex and from there to the contralateral auditory cortex. For example, after presentation of monosyllables in a study by Suzuki et al, 2.5 times more voxels were activated in the contralateral auditory cortex compared to the ipsilateral auditory cortex of the stimulated ear. 11 The combination of the assumptions described above results in a phenomenon called "the right ear advantage" (REA), meaning that in the majority of NH subjects, the right ear is most important for the perception and production of speech. Subsequent to a difference in hemispheric lateralisation between right-handed and left-handed people, there is a difference in the proportion of people that exhibit a REA between right-handed (79%) and left-handed people (68%). 7,14 Extensive research in NH subjects has demonstrated that the REA can be influenced by both bottom-up and top-down manipulations. 8,14 For example, in dichotic listening tests, the REA is augmented when the signal intensity in the right ear increases.
Conversely, when the signal intensity in the left ear increases, the REA shifts to a left ear advantage. Functional magnetic resonance imaging has demonstrated that the cerebral activation patterns also differ in the "right ear focus" and "left ear focus" situation, indicating the variance in processing of the auditory stimuli. Furthermore, EEG recording of event-related potentials (ERPs) has shown that the ERP latency from the right ear is shorter in the right ear focus situation, while it is longer in the left ear focus situation. 15 These examples demonstrate that the auditory processing and thus the perceived REA is actually affected by top-down manipulations. This led us to question whether hearing loss may affect the REA as well.
Although bilaterally deafened subjects benefit from bilateral CIs more than from a single CI with regard to speech perception in noise and localisation of sounds, currently the majority of deaf subjects is still implanted unilaterally mainly due to reimbursement issues. 16

Keypoints
• One hemisphere in the brain is of greater influence on several brain functions compared to the contralateral hemisphere. As such, a dominant left hemisphere is seen in 95-98% of right-handed and in 70-80% of left-handed normal hearing subjects for speech perception and speech production.
• Although the auditory cortex receives auditory input from both ears, it is most strongly stimulated by the contralateral ear.
• The combination of the assumptions described above results in a phenomenon called the right ear advantage, meaning that in the majority of normal hearing subjects the right ear is most important for the perception and production of speech.

| Study eligibility criteria
Studies reporting original data on the effect of side of cochlear implantation on postoperative hearing, speech and language outcomes were included. We were interested in unilaterally implanted patients; therefore, studies with (some) bilaterally implanted patients were excluded, when no subanalysis for unilateral patients was presented. The outcome measures for cochlear implant performance were clinically administered hearing, speech and language tests.

| Quality assessment
As shown in Table 2, we appraised the selected studies independently for directness of evidence (DoE) and risk of bias (RoB), using predefined criteria. DoE was scored by evaluation of population, intervention and outcome. Items were scored as satisfactory, partly satisfactory or unsatisfactory.
We assessed RoB by the use of several criteria. First, we examined the design of a study and, if prospective, whether blinding and randomization were performed. Second, we assessed the standardisation of outcome measurements. Standardisation was assessed as satisfactory when a validated test was used. Third, we assessed the handling of missing data and loss to follow-up. Fourth, we assessed whether follow-up was at set times or the timing differed per patient. Finally, we assessed handling of possible confounders. Disagreement was resolved by discussion and consensus.

| Data extraction and analysis
Two researchers (V.J.C.K. and T.C.D.) independently extracted descriptive data regarding the onset of hearing loss (prelingually or postlingually), side of implantation and hearing, speech and language outcomes from the selected studies.

| Search strategy and study selection
As shown in Figure 1, our search identified a total of 3551 articles, of which 2541 were unique. After screening on title and abstract, 30 articles were left for full-text screening. Reference checking yielded five additional possibly relevant articles provided that data could be obtained from the corresponding authors.
Two references were published conference abstracts, one was a letter to the editor and twelve articles were not eligible because of a non-corresponding population, intervention or outcome. Consequently, twenty articles were eligible for critical appraisal. Cochrane #1: one-sided:ti,ab or "one sided":ti,ab or "single sided":ti,ab or left*:ti,ab or right*:ti,ab or unilateral:ti,ab or independent:ti,ab or single-sided:ti,ab or side:ti,ab 53 #2: cochlear:ti,ab and implant*:ti,ab #3: #2 or "cochlear prosthes*":ti,ab or "auditory prosthes*":ti,ab or "auditory implant*":ti,ab

| Assessing quality of studies
The critical appraisal is presented in

| Data extraction
We were unable to pool data, because there was large clinical heterogeneity among these studies and measures of uncertainty were often not reported. Therefore, we provided a descriptive analysis.

| Study characteristics
Most of the studies were retrospective cohort studies, one prospective cohort study, 19 two cross-sectional studies 30 Table 4, three of five studies describing prelingual children with SNHL showed a significant REA for speech perception or speech production. 24

| Adults
As shown in Table 5, of the eight studies investigating the REA in adults with postlingual SNHL, four studies reported a significant REA for speech perception 19,21,25,28 and one for speech production, 31 while 3 other studies did not for speech perception. 1,2,27 In adults with prelingual SNHL, Kraaijenga et al found no significant difference in speech perception between left and right implantees. 1 Budenz et al showed that right ear implanted adults performed better on CNC words and phonemes, while the City University of New York tests for speech perception in quiet and noise did not differ significantly. 21 Similarly, Kamal et al observed a significant difference favouring the right CI in one of the three tests used. 25 Finally, Sharpe observed a significant REA for hearing in noise test scores after standardising the scores, yet not for the CVC word and phoneme scores. 28

| Handedness
Two studies performed a subanalysis on handedness. In concordance with their overall results, Morris et al found no difference between left ear and right ear implantees in a subgroup of right-handed patients. 2 Henkin et al 24 who found a significant REA in the whole patient group, also performed a subanalysis on the right-handed patients and found that the significantly better performance of right-    implanted children. 38 These results suggest that the left ear implanted children's brains adapted to unilateral auditory input in the left ear and consequently reafferentation took place to ensure sufficient input to the dominant left hemisphere.
Such altered patterns of input (increased ipsilateral and decreased contralateral processing of speech compared to NH subjects) have been demonstrated in adults with unilateral SNHL as well. 39,40 In these adults, an increase in ipsilateral activation of the auditory cortex was seen on the dominant left side. In children with unilateral SNHL, worse verbal and non-verbal performances with right-sided, unilateral SNHL compared to left-sided, unilateral SNHL have been described, 41,42 indicating that it is the right ear that adds the greatest contribution to speech perception.
In this systematic review, we provide an overview of studies that investigated the influence of side of implantation on hearing, speech and language outcomes after unilateral cochlear implantation. Four T A B L E 5 Study outcomes for subjects with postlingual onset of sensorineural hearing loss (SNHL) for several potential confounders such as type of implant device, follow-up and age at implantation. Therefore, this study was not included in this systematic review. To sum up results from these three studies, handedness does not seem to be an influential factor in the existence of the REA. This could be explained by the fact that left-handed patients predominantly have a left hemisphere dominance just as the majority of right-handed patients (70%-80% vs 95%-98%), therefore handedness did not greatly alter the effect of side of implantation.
As mentioned previously, the heterogeneity in the study populations is high, which is likely to have influenced the results. Furthermore, due to heterogeneity in outcome measures and a lack of reporting of measures of uncertainty, no pooled analysis could be performed. In addition, the used outcome measures may not be sufficiently specific to detect a REA. As described in the introduction, a REA is most evident in dichotic listening tests. These can, however, only be used in subjects who are able to receive auditory stimuli in both ears, which is evidently not possible for unilaterally implanted patients with symmetrical severe-to-profound SNHL.

| CONCLUSION
This systematic review investigated the effect of side of implantation in subjects with symmetrical SNHL after receiving a CI. The results of this review, though the level of evidence of included studies was low, revealed evidence for a REA in prelingually deafened children as well as postlingually deafened adults. Moreover, none of the studies identified a left ear advantage. In view of these results, we suggest that right ear implantation may be beneficial over left ear implantation in case of symmetrical SNHL and in the absence of other prognostic factors. Furthermore, this will be the practical side for most (right-handed) patients.

CONFLI CT OF INTEREST
The authors do not have any competing interests.