Cochlear implantation as a treatment for single-sided deafness and asymmetric hearing loss: a randomized controlled evaluation of cost-utility

Study design and setting

This prospective multicentre national study combines two major steps:

– a prospective, descriptive observational cohort study, with a 6-month follow-up for SSD/AHL adult subjects treated by CROS, SSD/AHL subjects treated by an implanted bone conduction device (Baha or Ponto), and adult patients who decline all the options.

– and an open-label randomised controlled clinical trial for adults with SSD or AHL after failure of both CROS and BC systems, in two parallel groups: Observation for 6 months versus Cochlear implantation.

Participants are recruited in 7 tertiary referral centres in France. Once the inclusion/criteria are checked and the consent from the subject is obtained by one of the investigators, a reference evaluation is made for quality of life, binaural hearing performances (i.e. speech recognition in noise and horizontal localisation) and tinnitus severity. Then, CROS hearing aids and a BC device on a headband are successively tried by the subjects for three weeks each. The evaluation of quality of life, binaural hearing abilities, and tinnitus severity is repeated after the first trial (CROS) and the second trial (BC). At this point, the subject may choose the treatment he/she is willing to continue. If one of the trials (CROS or BC) is considered to be successful and the subject is willing to continue with the corresponding device, it is then provided and a new evaluation is performed 6 months later. If none of the trials is considered to be successful and the subject does not want any other treatment, then only a new evaluation is performed 6 months later. If none of the trials is considered to be successful and the subject wants treatment by cochlear implantation, the second major step of the study begins with the randomisation procedure.

Subjects who chose a cochlear implant are thus randomised between two arms: in the first one (“CI” arm), a cochlear implantation is performed within a month and a new evaluation on quality of life, binaural hearing performances, and tinnitus severity is planned 6 months after CI activation. Subjects of the second arm (“observation” arm) simply undergo the new evaluation 6 months after randomisation, without any intercurrent treatment for SSD/AHL. For obvious ethical reasons, subjects of the second arm benefit from a cochlear implantation after this 6-month evaluation.

Data from all the subjects included will be anonymised, collected and analysed. If a serious adverse event does not allow the end of the protocol to be reached and/or if a subject withdraws his/her consent, the data which were collected until this event will also be taken into account in the final analysis. If a serious adverse event occurs, it will be reported to the coordinating centre (Toulouse University Hospital) and to the “Agence Nationale de Sécurité du Médicament” (ANSM, National Agency of Drug Safety).

Inclusion criteria

Age eighteen years or older.

French speaking.

French social security affiliation.

Ability and willingness to participate in all the assessments, with written consent to the protocol.

Poorer ear: severe-to-profound hearing loss measured using pure tone audiometry with pure-tone average (0.5, 1, 2, 4 kHz) ≥ 70 dB, confirmed by auditory brainstem responses, with minimal benefit of a conventional hearing aid. In order to account for the heterogeneity of this population, no selection was applied on the duration of deafness or the severity of the tinnitus.

Better ear: normal hearing thresholds (≤20 dB HL) to moderate hearing loss (PTA ≤ 60 dB HL).

Exclusion criteria

Retrocochlear pathology (vestibular schwannoma, severe central auditory processing disorder).

Major cochlear ossification or malformation which could prevent the full insertion of a cochlear implant, controlled on CT scan and MRI of temporal bones.

Subjects under juridical protections or tutelage measure.

Neurological or psychiatric disorder.

Previous treatment with CROS, BC or CI.

Medical condition that contraindicates CI surgery.

Randomisation procedure

Randomisation of CI versus observation was based on a 1:1 ratio and was stratified per centre. The allocation sequence was randomly generated by a computer (Stata SE 11.2, ralloc procedure), providing a sequence of treatments randomly permuted in blocks of varying size [2, 4, 6]. There was no blinding procedure in this open-label trial; however randomisation was centralised at the Toulouse University Hospital and managed by the project manager and clinical research associates. The allocation sequence was unknown to the investigators who enrolled participants.

Population and sample size

The sample size was calculated based on the expected number of failures of CROS hearing aids, BC trials and on the expected improvement of quality of life in cochlear implanted subjects with SSD/AHL. In a pilot study performed in our centre (unpublished), 40 to 50% of SSD/AHL subjects declined both CROS and BC devices after a three-week trial. This rate is consistent with previous studies performed on BC implantation after a headband trial (45% in Desmet et al. [33]). An improvement in quality of life is expected in these subjects after cochlear implantation, but it is not as significant as in patients with bilateral severe-to-profound deafness. A recent meta-analysis performed on traditional CI candidates with severe-to-profound deafness demonstrated an improvement of 1.05 standard deviations. To detect an improvement of 0.8 standard deviations with an alpha risk of 5% and an 80% power, 50 subjects (25 per arm) are necessary. Altogether, and to compensate the risk of dropouts, 150 subjects are included in this study.


The purpose of this study is to assess the cost-utility ratio of CI in subjects with SSD/AHL after failure of more conventional treatments. Therefore, every subject will successively try CROS hearing aids and a BC device placed on a headband, for three weeks each. The baseline assessment of each trial relies on auditory measures (speech recognition in noise and horizontal localisation) and quality of life questionnaires. Once each trial has been performed, the subject will decide which treatment he/she opts for. This choice is mainly based on the subjective feelings of the subject after the two trials, but also guided by the auditory outcomes obtained with each device, and the counselling of the physician. The programme provides all costs directly related to the treatment, i.e. the cost of the medical device and the following fitting sessions.

The information regarding the results which may be expected from a cochlear implant is critical. A CI is the only device which cannot be herein tried before the decision is made and its choice intrinsically relies on hedging a bet. Across the seven centres, this information is standardised through several elements indicated to the subjects willing to try a CI, based on the literature on cochlear implantation outcomes in SSD/AHL [









  • CI may restore binaural hearing to some extent but with a significant inter-individual variability;

  • its effect is more consistent on ipsilateral tinnitus when associated with unilateral profound deafness;

  • The impact of CI in this indication has mainly been studied and demonstrated in subjects where the duration of profound deafness was less than 10 years.

Once this information has been delivered, each subject makes a final choice among four options: abstention, CROS hearing aids, Bone-Anchored hearing aid (Baha or Ponto), or CI. Subjects who chose CI are randomised between two arms (“CI” arm versus “observation” arm) and re-assessed and compared after 6 months their choice has been implemented, and the results of the two arms are compared. Subjects who chose abstention, CROS or BAHA are re-assessed 6 months after their choice has been implemented, and the results of each group are described.


Subjects in this group who do not experience any benefit from the CROS hearing aid trial or from the BAHA trial, and choose no specific treatment for SSD/AHL.

CROS hearing aids

In CROS systems, a remote microphone which is placed on the poorer ear receives the surrounding incoming signal and transmits it to the second hearing aid on the better ear, using a wireless link. The second hearing aid delivers the signal using air conduction and can also apply an additional amplification in case of hearing loss in the better ear.

A Phonak CROS system will be used for the trials and in case the subject eventually opts for this type of rehabilitation (Phonak AG, Stäfa, Switzerland). The hearing aids will be fitted by an experienced audiologist, using the fitting software provided by the company (Phonak Target™ 3.0).

Bone conduction device (BAHA)

Bone-anchored hearing aids transmit the auditory signal received by the processor placed behind the poor ear to the better ear through bone conduction of the skull. The programme does not provide any additional hearing aid in the better ear if the subject has AHL.

Baha BP 110® (Cochlear Ltd) and Ponto® (Oticon) will be used on a headband for the trial period and placed on the corresponding abutment after a surgical procedure if the subject chooses this option. The surgery consists of the placement of the implant which aims at being osteo-integrated, on which an abutment is screwed. The processor may be used for 3 to 6 weeks following the surgery according to wound healing. It is then fitted by an experienced audiologist following the company’s guidelines.

Cochlear implant

A surgical procedure is performed to allow the placement of the receptor under the temporal muscle and the introduction of the electrode array into the scala tympani via the round window or a cochleostomy. According to the company, the electrode array supports 12 to 22 intra-cochlear electrodes which aim at stimulating the first neurons of the auditory nerve. Four cochlear implant companies are involved in this study: Advanced Bionics, Cochlear, MedEl and Oticon. The choice of the brand will depend on the physician in charge of the patient. Approximately four weeks after surgery, the external processor can be activated and the patient will therefore receive electric auditory stimulation of the implanted ear.

Primary outcome measure

The primary outcome measure will be applied to the two randomised arms to determine the incremental cost-utility ratio (ICUR) of cochlear implantation versus initial observation at 6 months. The utility is defined in this study as survival weighted by the quality of life of patients at 6 months (QALY). Quality of life will be measured using a generic questionnaire, the EuroQol-5D-3 L scale. This instrument is recommended by the national French agency (Haute Autorité de Santé) for cost-utility studies [36].

The EQ-5D-3 L is a self-administered, generic and multidimensional questionnaire [37]. It has two components, a descriptive component and a visual analogue scale (EQ-VAS). The descriptive component is composed of 5 dimensions described by three levels that define 243 health states. The EQ-5D-3 L is translated and validated into the French language and it has a utility function calculated on the basis of known preferences of the French population [38]. The utility-preference approach of the EQ-5D-3 L provides a cardinal measuring instrument available to calculate the cost-utility ratios.

Healthcare costs will be assessed from the French health insurance perspective [39, 40]. Direct medical and non-medical costs will be included in this study. Costs related to productivity loss will also be analysed. Direct medical costs correspond to hospitalisation costs, outpatient costs (i.e. visits and medical acts, paramedical acts), medication and medical device costs, especially the cost of the CI. Non-medical costs include transportation costs. Data linked to the number of days the patient has missed work will also be gathered from the French health insurance databases. Costs will be estimated by multiplying the number of units used for each resource with the corresponding unit cost.

Consumption of healthcare resources will be retrospectively gathered from the French Social Health Insurance databases, using a bottom-up approach. Administrative data corresponding to the name, surname, date of birth, place of living and gender were recorded for patients included in the study.

Inpatient stays will be valued using the French Disease Related Groups (DRGs). Outpatient care, which includes visits, medical and paramedical acts, will be valued using the tariffs reimbursed by the French health insurance. Visits and paramedical acts will be valued using the French General Nomenclature of Professional Acts. Medical acts will be valued using the French Common Classification of Medical Acts, except for laboratory tests for which valuation will be based on the Nomenclature of Biological Acts. For all these fees, we will apply the corresponding reimbursement rate and we will subtract, if necessary, the medical deductible that is due by the patient and not reimbursed by the French Social Health Insurance (FSHI).

A cost-utility analysis will be performed. This study will establish a link between costs and medical consequences, expressed in QALY gained, at 6 months between the two care management strategies. The ICUR between SSD/AHL patients treated by CI and those that are only followed-up, will be calculated as follows (Drummond et al., 2005):

$$ ICUR=frac{Delta C}{Delta U}=frac{Mean cost left( CIright)- Mean cost (abstention)}{Mean QALY (CI)- Mean QALY (abstention)} $$

Where ΔC and ΔU were increments of costs and utilities, respectively.

A Budget Impact Analysis (BIA) will be implemented to measure the net costs to the social health insurance of the care management of SSD/AHL patients, taking into account all positive and negative variations, the use of health resources that could result from the CI [41].

The BIA, which will be conducted using international and French guidelines, will include the predictable variation of context elements (e.g. demographic and epidemiological changes) [41, 42]. The social health insurance perspective will be taken and the time horizon will be 3 years.

The analysis will include three phases:

– An inventory of the care management of SSD/AHL patients,

– A care management model of these patients after the introduction of the new strategy (i.e. CI) to their care,

– The estimated cost for health insurance for these situations and the difference between the two strategies, to estimate the financial impact of the introduction of the new strategy in the treatment of SSD/AHL patients.

The introduction of the strategy based on the CI can change their own cost, but also have an impact on the size of the population (target and reached population), patterns of care performed prior to and after, the unit costs of resources mobilised in the context of strategies. The BIA should take these factors into account.

Secondary outcome measures

All participants are asked to complete several quality of life questionnaires and hearing tests at inclusion and 6 months after their choice of the treatment (abstention, CROS hearing aids, BAHA, cochlear implantation). Pre/post treatment analyses will be carried out in each of these groups. Comparisons will be made in the group of subjects who chose a CI, randomised between immediate CI and initial observation.

Quality of life

As described above, a generic evaluation of quality of life is performed using the EuroQol-5D-3 L scale. Hearing-specific quality of life questionnaires will also be administered.

The Nijmegen Cochlear implant questionnaire (NCIQ) contains 60 questions exploring 6 areas concerned with the quality of auditory perception (basic perception, complex perception, speech production, self-esteem, social activities and interactions) [43]. Indeed, the perception of basic environmental sounds such as bells or footsteps is assessed as well as oral communication with new contacts. Each item is formulated as a statement with a 5 point response scale to indicate the degree to which this statement is judged true (from “never”:1 to “always”:5). An extra point can be chosen if none of the points fits.

An evaluation of the discomfort related to the possible tinnitus associated with the deafness will rely on a visual analogue scale (VAS) ranging from 0 to 10. This scale is presented as a 17 cm plastic ruler with a vertical arrow on one side and a graduated scale on the other side (0 to 10 cm). The subject first indicates the level of annoyance generated by the tinnitus on the vertical arrow using a cursor and the corresponding numeric value is reported by the evaluator. Then, the intensity of tinnitus is assessed using another ruler with the same dimensions.

Speech understanding in noise

Speech recognition in competing noise is measured using the French Matrix test [44] in sound field (IAC 120A-1 sound booth). The French Matrix test is a closed-set sentence test that uses 50 well-known words in French. Each sentence has the same syntactic structure: name – verb – number – object – colour, for example: “Felix draws six blue bikes”. The number of combinations of five words is large enough to eliminate memory-based responses. Speech and noise signals are generated from an IBM PC running the OMA software ( and presented via loudspeakers and amplifier (Studio Lab, SLB sat 200). Speech signal is presented at a fixed level of 65 dB SPL and the level of competing noise is adjusted using the adaptive procedure described by Jansen et al. to obtain the signal-to-noise ratio in dB for 50% correct word recognition (SNR50).

SNR50 is obtained in three different spatial configurations: one with speech and noise presented from a single loudspeaker in front of the subject at 0° and two conditions with speech and noise presented from separate loudspeakers at 60° to the left and right of the subject. The choice of a spatial configuration (− 60°, 0° and + 60°) was selected based on previous results [45] showing that when such positions are used, the head-shadow effect is reduced while the binaural unmasking is maximised compared to a ± 90° configuration. The “dichotic” condition is defined as speech presented to the poorer hearing ear and the noise to the contralateral, normal hearing ear; the diotic condition as both the signal and the noise presented from the loudspeaker located in front of the subject; and the reverse dichotic condition with speech presented to the normal hearing ear and noise to the poorer ear. Subjects will be asked to repeat any word which is heard.

Auditory horizontal localisation

Horizontal localisation ability is assessed using an array of 7 horizontal loudspeakers and an amplifier (Studio Lab, SLB sat 200) located at intervals of 30 degrees from – 90 degrees to 90 degrees in a frontal semicircle diameter of 1.2 m at the subject’s head level. The stimulus comprises two 150 msec white gaussian noises from 20 Hz to 20 kHz with a 0.05 msec ramp. A silence of 150 msec is inserted between the two noises for total stimulus duration of 450 msec. This stimulus is similar to the one used by Slattery and Middlebrooks [46] and is presented 63 times (9 presentations per loudspeaker) with a period of 2 s silence between each presentation to allow the subjects to indicate orally the loudspeaker location. Subjects are asked to not move their head during the experiment: this was controlled by the examiner. Localisation ability measurements are the rate of correct localisation and the root mean square error.

Statistical methods, data reporting and analysis

Once the validity of the database is double-checked, descriptive analyses will be carried out for each treatment option (observation, CROS hearing aids, bone conduction device, cochlear implant). Quantitative variables will be described as means, standard deviations, and percentile distribution. The number and proportion of subjects will be used to describe categorical data.

Regarding the randomised controlled trial phase comparing CI versus observation, the analysis will be conducted on all subjects who were randomised on an intention-to-treat basis. For each outcome (quality of life, hearing performances indicators), the value measured at 6 months will be compared between the two groups using bivariate analyses (Student’s T test or Wilcoxon rank sum tests) and linear regressions adjusted for the baseline value.

For each intervention, we will assess the associations between changes in hearing performances and changes in quality of life. Data will be presented according to the Consolidated Standards of Reporting Trials (CONSORT) Statement [47, 48].

In order to test the robustness of the ICUR, deterministic and probabilistic sensitivity analyses will be conducted.

As part of the sensitivity analysis, we will determine the robustness of the results by testing the impact on the outcome of the variation of different cost and utility parameters [40]. Probabilistic sensitivity analysis, performed using the non-parametric bootstrap method, identified the uncertainty around the ICUR by estimating its confidence interval [49]. Moreover, a Cost-Utility Acceptability Curve (CUAC) will be built to summarise the impact of uncertainty on the ICUR in relation to possible values of the cost-effectiveness threshold [50].

Syndicated From