A retrospective, observational, and multicenter study was conducted including patients over 18 years of age diagnosed with BVP of more than 1 year of duration, who were recruited from two tertiary referral centers. The two hospitals were Clínica Universidad de Navarra (Pamplona and Madrid, Spain) and Hospital Universitario Insular Materno-Infantil de Gran Canaria (Las Palmas de Gran Canaria, Spain).

The study received ethical approval from the competent authorities and their respective local ethics committees. Written informed consent was obtained from all participants. The study was conducted in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards, where applicable. All patients provided written informed consent prior to participation, and all procedures involving human participants adhered to the ethical standards of the institutional research committees.

Inclusion criteria contemplated patients over 18 years of age diagnosed with bilateral vestibulopathy of at least one year of evolution, according to the diagnostic criteria of the Bárány Society Consensus (1).

Exclusion criteria comprised patients with inner ear malformations or other structural abnormalities, unilateral vestibulopathy, or with significant psychiatric comorbidities likely to impair informed consent, data reliability, or adherence to the study protocol. Patients who refused or were unable to provide written informed consent or to participate in the study were also excluded.

All patients underwent a comprehensive physical examination, including otoscopy and a detailed otoneurological assessment using videonystagmography (VNG) (VideoFrenzel, Interacoustics VF505m, Assens, Denmark). The audiovestibular evaluation comprised pure-tone audiometry (PTA) (AC40, Interacoustics), vestibular evoked myogenic potentials (VEMP) (Eclipse, Interacoustics, Assens, Denmark), the video Head Impulse Test (vHIT) (ICS Impulse, GN Otometrics®, Natus Medical, Denmark) and dynamic posturography (EquiTest, NeuroCom International Inc., Clackamas, OR, USA). When clinically indicated, caloric testing (VisualEyes™, Interacoustics, Denmark) and dynamic visual acuity assessment (Balance Master System, NeuroCom International Inc., Clackamas, OR, USA) were also performed. All patients underwent audiological and vestibular testing on the same day of the diagnosis. All patients underwent audiological and vestibular testing on the same day of the diagnosis. For all audiovestibular analyses, data were evaluated and reported in an ear-specific manner. Comparisons were restricted to anatomically and physiologically equivalent measures only, that is, ipsilateral ear-to-ear comparisons across etiologies (right-to-right and left-to-left), and within the same test modality and end organ. No cross-lateral (right vs. left) or cross-parameter comparisons (e.g., between different semicircular canals or between cVEMP and oVEMP measures) were performed. This analytical approach was chosen to preserve clinical interpretability and to reflect real-world decision-making in implant candidacy.

Clinical and demographic data were collected for all participants, including sex, age at diagnosis, etiology and disease evolution. Functional symptoms were systematically documented, including the presence or absence of imbalance, oscillopsia, vertigo spells during last year, and falls during daily activities. This information was used to characterize the clinical profile of each patient and to explore potential associations between etiology, disease evolution, and symptom in subjects potentially eligible for vestibular implantation.

Eligible participants were adults (≥18 years) with a diagnosis of bilateral vestibulopathy following the Bárány Society research criteria (1) of at least one year’s duration. The implanted ear had to show absent cVEMP and oVEMP responses and demonstrate optimal anatomical conditions for electrode insertion on CT scans and MRI (patent vestibule and intact vestibular nerve). Participants also needed to be able to undergo surgery and comply with a personalized rehabilitation program. From an auditory standpoint, for the vestibular-only device, candidates were considered those presenting a moderate to severe hearing loss (PTA 40–70 dB) in the ear to be implanted. For the combined cochleo-vestibular device, eligibility required a severe-to-profound hearing loss in the implanted ear and a speech recognition score below 50% at 65 dB on the Cárdenas and Marrero (22) disyllabic word test, measured under best hearing-aided conditions.

Exclusion criteria for implantation, even in patients fulfilling functional audiovestibular criteria, included inner ear ossification or malformations preventing full electrode insertion, middle ear disease or conductive hearing loss, retrocochlear pathology, and central causes of hearing or balance impairment. Central nervous system conditions comprised cerebellar ataxias as CANVAS, Parkinson’s disease, brainstem or cerebellar lesions, and other vascular or neurodegenerative disorders affecting gait or oculomotor control, as well as downbeat nystagmus syndrome, all identified through detailed neuro-otological and neurological examination and neuroimaging when appropriate. Patients with persistent postural–perceptual dizziness (PPPD), visually induced or functional vertigo, oculomotor abnormalities, peripheral neuropathies, unilateral vestibular deficits, or intoxication due to alcohol or drugs were also excluded, given the central or non-vestibular contribution to symptoms. Additional exclusion criteria included chronic depression, dementia, or other cognitive disorders likely to interfere with correct device use or outcome assessment; active psychiatric disease; poor compliance or unrealistic expectations regarding potential benefit; pregnancy or breastfeeding; participation in other clinical trials within the previous 30 days; or explicit patient refusal to participate.

When referring to the cochleo-vestibular implant used in this study, we employed the BionicVEST 1 (Cochlear Ltd., Sydney, Australia) model, which consists of two main components: an internal and an external unit. The internal component comprises a receiving coil, an intracochlear electrode array with 19 contacts configured according to the CI512 Contour Advanced (Cochlear Ltd., Sydney, Australia) perimodiolar design, a vestibular electrode array with three full-band contacts spaced 0.2 mm apart, and a reference ball electrode. In contrast, the vestibular-only implant, called BIONIC-VEST 2 (Cochlear, Sydney, Australia), consists solely of the vestibular electrode array and the reference electrode, without the intracochlear array.

The external component of both corresponds to the commercially available Nucleus 6 or 7 sound processor, which is conventionally positioned beneath the periosteum and oriented toward the zygomatic arch.

Stimulation parameters consisted of generating balanced biphasic pulse trains. The pulse width was set to 25 μs, with a stimulation frequency of 900 pulses per second. Monopolar stimulation was applied using the ball and plate electrodes as reference (MP1+2 configuration). The modulation of stimulation varied according to the target structure: for cochlear stimulation, the amplitude of the stimulus was modulated between the threshold (T) and comfort (C) levels, defining the dynamic range, whereas otolithic stimulation was delivered at a constant current level.

Continuous variables were tested for normality using the Shapiro–Wilk test. Normally distributed data are presented as mean ± standard deviation (SD), whereas non-normally distributed variables, such as PR indices, are reported as median [interquartile range, IQR]. As most continuous parameters did not meet normality assumptions, comparisons across etiological groups and implant-candidacy categories were performed using the Kruskal–Wallis test. Pairwise comparisons were conducted using Dunn’s post hoc test with Bonferroni adjustment; therefore, p-values displayed for post hoc pairwise comparisons are adjusted for multiple testing. Categorical variables and symptom frequencies were analyzed using the Chi-square test or Fisher’s exact test, as appropriate. Again, when multiple categorical comparisons were performed, Bonferroni correction was applied, and the reported p-values for these multiple comparisons are adjusted.

The association between oscillopsia and vestibular or functional parameters was assessed using binary univariate logistic regression, with oscillopsia (present/absent) as the dependent variable. Semicircular canal VOR gains (SSC, LSC, and PSC), PR indices, posturographic performance (SOT), visual acuity, age, and disease duration were each evaluated as independent predictors in separate models. Results are expressed as odds ratios (OR) with 95% confidence intervals (CI), and statistical significance was assessed at a two-sided alpha level of 0.05.

All statistical analyses were performed using GraphPad Prism v8.0.1 (GraphPad Software Inc., San Diego, CA, USA).

Of the 170 patients initially diagnosed with BVP, 51 (30%) were classified as probable cases and 119 (70%) fulfilled the Bárány Society criteria for definite BVP (1), constituting the final study cohort. The demographic characteristics of the patients included in the study are summarized in Table 1.

All patients (n = 119, 100%) presented with chronic imbalance, fulfilling the diagnostic criteria for BVP, while oscillopsia—a hallmark symptom of the disease—was reported by 58 participants (48.74%). Additionally, 66 patients (55.46%) experienced episodic vertigo and 27 (22.69%) had a documented history of falls.

From a general perspective, the audiometric evaluation revealed slightly poorer hearing thresholds in the right ear. Although the right ear exhibited a higher mean PTA, both showed a similar distribution of severe-to-profound hearing loss range: 32 patients (26.89%) on the right ear and 30 patients (25.21%) on the left ear.

vHIT results showed that in general, the lowest gains were observed in the posterior semicircular canal (PSC), followed by the lateral semicircular canal (LSC) and superior semicircular canal (SSC). Notably, 18 patients (15.13%) who exhibited LSC gain values above 0.6 but presented a clinical presentation consistent with BVP, showed complete vestibular areflexia following ice–water irrigation in both ears in the caloric test, with a mean SPV of 0.42 ± 0.54°/s. Finally, an intermediate value was observed in both ears for the PR index, which did not differ statistically.

cVEMPs were absent in 60 patients (50.42%), while oVEMPs were absent in 63 patients (52.94%). Finally, posturography outcomes based on the SOT and LOS, as well as dynamic visual acuity were consistently found within pathological ranges. A more detailed description of these results is provided in Supplementary Table, in which no statistically significant differences were observed in any variable when compared with the contralateral side.

Symptom analysis across etiological groups revealed distinct clinical patterns. While unsteadiness was a universal finding among all participants, the distribution of other symptoms varied notably between etiologies. Vertigo episodes were most frequent in patients with Ménière’s disease, consistent with its episodic course, whereas oscillopsia was predominantly reported in subjects with vestibulotoxic, post-infectious, and cerebellar or ataxic origins, reflecting a more chronic vestibular impairment. Conversely, the occurrence of falls was disproportionately higher in individuals with cerebellar ataxia or CANVAS, suggesting a more severe impact on postural control in this subgroup. Overall, these results emphasize the heterogeneous clinical presentation of bilateral vestibulopathy, where symptom burden and compensation profiles differ depending on the underlying cause. It can be inferred in Table 2.

As illustrated in Figure 1, despite the absence of a generalized difference across groups in the Kruskal–Wallis (Table 3), mean hearing thresholds were broadly comparable among etiologies. The post-infectious subgroup exhibited the highest PTA values, whereas idiopathic and especially CANVAS/ataxia patients tended to show lower thresholds. Considerable variability was observed in the idiopathic and iatrogenic groups, as reflected by the wide dispersion of their boxplots. Overall, both ears demonstrated similar distributions of hearing loss severity, predominantly within the moderate-to-severe range.

As illustrated in Figure 2 and summarized in Table 3, significant inter-etiological differences were identified in VOR gain across all semicircular canals (p < 0.001). Idiopathic patients showed the highest gains, followed by those with post-traumatic lesions and Ménière’s disease, whereas CANVAS and vestibulotoxic etiologies presented the most pronounced deficits. Post-infectious cases occupied an intermediate position.

The LSC showed the largest intergroup separation, confirming a consistent pattern of functional decline across canal planes. Bilateral symmetry of results further supports the uniformity of vestibular damage within each etiology. Even more, in LSC, despite PR values did not differ significantly between groups (p > 0.05), again CANVAS and vestibulotoxic etiologies presented the most pronounced deficits in terms of a higher level of saccades disorganization with higher scores in the index.

Regarding VEMPS, the rate of absent responses differed markedly across etiologies: CANVAS (cVEMP 75.00%, oVEMP 87.50%) and post-traumatic (60.00, 80.00%) exhibited the highest absence rates, followed by iatrogenic/vestibulotoxic (63.63, 59.09%). In contrast, absence rates were lower in idiopathic (45.71, 42.86%), Ménière’s disease (40.62, 37.50%), and post-infectious (47.06, 41.17%) patients. Among participants with measurable responses, as shown in Table 3, Figure 3, IAAR values revealed a consistent functional hierarchy: idiopathic and Ménière’s disease exhibited the highest amplitudes, while CANVAS, vestibulotoxic, and post-traumatic etiologies displayed markedly lower IAAR values. Post-hoc comparisons confirmed that idiopathic and Ménière’s patients had statistically significant higher IAAR than CANVAS and vestibulotoxic groups for both cervical and ocular recordings, despite presenting a clearly lower percentage of absent responses.

Finally, as shown in Figure 4 and summarized in Table 3, dynamic posturography revealed also significant inter-etiological differences for both SOT and LOS (p = 0.007 and p = 0.010, respectively). Idiopathic, post-traumatic and Ménière’s disease patients achieved the highest scores, indicating better postural stability, whereas post-infectious and CANVAS groups showed the lowest performance. Iatrogenic etiologies occupied an intermediate position, with broad variability and without clear pairwise separation.

Binary univariate logistic regression analyses identified several vestibular predictors significantly associated with the presence of oscillopsia (Table 4). Reduced VOR gain in the LSC was significantly associated with oscillopsia on both sides (right LSC: OR 0.185, p = 0.036; left LSC: OR 0.149, p = 0.035). Likewise, reduced PSC gain was also significantly associated with oscillopsia bilaterally (right PSC: OR 0.167, p = 0.043; left PSC: OR 0.182, p = 0.047).

In addition, the PR index of the right LSC emerged as a significant positive predictor of oscillopsia (OR 1.045, p = 0.010), whereas the left PR index showed only a trend toward significance (OR 1.009, p = 0.054). In contrast, SSC gains, SPV results of the caloric test, posturographic performance (SOT score), visual acuity and clinical variables as age, disease duration, were not significantly associated with oscillopsia (p > 0.05 for all).

Of the 119 patients included in the study, those who failed to achieve sufficient improvement after conventional vestibular rehabilitation were evaluated for potential implantable therapies. After excluding patients with central or neurodegenerative disorders—such as CANVAS—from implant eligibility, 28 patients (23.53%) were considered suitable candidates for implantable devices: 14 (11.76%) for a cochleo-vestibular implant (Bionic-VEST 1) and 14 (11.76%) for a vestibular implant (Bionic-VEST 2). The remaining 91 patients (76.47%) were classified as non-candidates. Group characteristics according to etiology and clinical features are summarized in Table 5.

Among cochleo-vestibular implant candidates, Ménière’s disease was the most frequent etiology, followed by idiopathic and post-infectious causes. In contrast, idiopathic and Ménière’s disease predominated among vestibular implant candidates, with iatrogenic and post-infectious etiologies less commonly represented. Post-traumatic cases were infrequent across implant groups.

Unsteadiness was present in all patients regardless of candidacy status. Oscillopsia was more frequently observed among implant candidates—particularly in the cochleo-vestibular group—compared with non-candidates, showing a trend toward statistical significance (p = 0.094). Falls were more common among cochleo-vestibular implant candidates, whereas vertigo spells were predominantly reported in non-candidates.

From an audiometric perspective, significant intergroup differences were observed (p < 0.001). As expected, patients selected for the cochleo-vestibular implant exhibited markedly poorer hearing thresholds in both ears (right PTA 104.64 dB, left PTA 89.06 dB), consistent with the combined auditory and vestibular indication of this device. In contrast, candidates for the vestibular implant showed relatively preserved cochlear function, with mean PTA values (46.91 dB right, 47.78 dB left) comparable to those of non-candidates (56.38 dB right, 55.11 dB left). These findings are summarized in Table 6, Figure 5.

Global vHIT analysis revealed significant intergroup differences across all semicircular canal planes (Table 6, Figure 6). The cochleo-vestibular candidate group demonstrated the most pronounced impairment, with significantly reduced gains in all canals and markedly elevated PR indices, indicating severe bilateral vestibular dysfunction with substantial saccadic disorganization. The vestibular implant candidate group showed slightly higher, yet still pathological, canal gains, suggesting partial residual function or compensatory mechanisms. Non-candidates exhibited comparatively higher gains and lower PR values, consistent with milder vestibular impairment. Overall, Kruskal–Wallis testing with post-hoc comparisons confirmed significant differences between groups for all canal gains (p < 0.05). In addition, among the 18 patients who underwent caloric testing, 16 were classified as non-candidates for implantation, with a mean SPV of 0.39 ± 0.54°/s. Of the remaining two patients, one, with an SPV of 1.10°/s, was considered a candidate for a cochleo-vestibular implant, while the other, with an SPV of 0.50°/s, was selected for a vestibular implant. Nevertheless, these differences did not reach statistical significance (p = 0.329).

Because absent VEMP responses constituted a prerequisite for implant candidacy, no intergroup comparison of VEMP outcomes was performed in this section, in contrast to Section 3.3, as all implant candidates lacked measurable responses whereas non-candidates did not. Despite the absence of statistically significant differences between groups for posturographic measures (SOT p = 0.424; LOS p = 0.709), the cochleo-vestibular group tended to exhibit lower SOT scores. Interestingly, the non-candidate group—despite moderate SOT performance—showed the lowest LOS values, suggesting less efficient dynamic postural strategies. These results are detailed in Table 6 and illustrated in Figure 7.

This study has several limitations that should be acknowledged. First, its retrospective design may introduce selection and information biases, and causality cannot be directly inferred from the associations reported. Second, although our sample of definite BVP cases was relatively large, several etiologies remained modest in size, reducing the statistical power for subgroup analyses and widening confidence intervals in those categories. Third, we relied on standard clinical metrics for oscillopsia, based on patient-reported experience. Since oscillopsia is intrinsically a dynamic symptom, the lack of treadmill-based measurements or high-velocity gaze-stabilization testing may underestimate its real functional. Even more, the measurement of some vestibular tests as the lack of quantitative caloric SPV values, as responses were analyzed in a binary manner based on the presence or absence of nystagmus during ice-water stimulation, which does not allow strict application of the Bárány bithermal SPV criterion. Finally, the cross-sectional nature of our analysis does not allow evaluation of longitudinal changes in compensation or auditory–vestibular evolution, which may influence implant candidacy over time.