The n e w e ng l a n d j o u r na l of m e dic i n e

Original Article

Posture, Gait, Quality of Life, and Hearing

with a Vestibular Implant
M.R. Chow, A.I. Ayiotis, D.P. Schoo, Y. Gimmon, K.E. Lane, B.J. Morris,
M.A. Rahman, N.S. Valentin, P.J. Boutros, S.P. Bowditch, B.K. Ward, D.Q. Sun,
C. Treviño Guajardo, M.C. Schubert, J.P. Carey, and C.C. Della Santina​​

A BS T R AC T

BACKGROUND
Bilateral vestibular hypofunction is associated with chronic disequilibrium, postural The authors’ full names, academic de-
instability, and unsteady gait owing to failure of vestibular reflexes that stabilize the grees, and affiliations are listed in the Ap-
pendix. Address reprint requests to Dr.
eyes, head, and body. A vestibular implant may be effective in alleviating symptoms. Della Santina at 720 Rutland Ave., Ross
METHODS Bldg., Rm. 826, Baltimore, MD 21205, or
at ­cds@​­jhmi​.­edu.
Persons who had had ototoxic (7 participants) or idiopathic (1 participant) bilateral
vestibular hypofunction for 2 to 23 years underwent unilateral implantation of a Ms. Chow, Ms. Ayiotis, and Drs. Schoo
and Gimmon contributed equally to this
prosthesis that electrically stimulates the three semicircular canal branches of the article.
vestibular nerve. Clinical outcomes included the score on the Bruininks–Oseretsky
N Engl J Med 2021;384:521-32.
Test of Motor Proficiency balance subtest (range, 0 to 36, with higher scores indicating
DOI: 10.1056/NEJMoa2020457
better balance), time to failure on the modified Romberg test (range, 0 to 30 seconds), Copyright © 2021 Massachusetts Medical Society.
score on the Dynamic Gait Index (range, 0 to 24, with higher scores indicating better
gait performance), time needed to complete the Timed Up and Go test, gait speed,
pure-tone auditory detection thresholds, speech discrimination scores, and quality of
life. We compared participants’ results at baseline (before implantation) with those at
6 months (8 participants) and at 1 year (6 participants) with the device set in its usual
treatment mode (varying stimulus pulse rate and amplitude to represent rotational
head motion) and in a placebo mode (holding pulse rate and amplitude constant).
RESULTS
The median scores at baseline and at 6 months on the Bruininks–Oseretsky test were
17.5 and 21.0, respectively (median within-participant difference, 5.5 points; 95%
confidence interval [CI], 0 to 10.0); the median times on the modified Romberg test
were 3.6 seconds and 8.3 seconds (difference, 5.1; 95% CI, 1.5 to 27.6); the median
scores on the Dynamic Gait Index were 12.5 and 22.5 (difference, 10.5 points; 95%
CI, 1.5 to 12.0); the median times on the Timed Up and Go test were 11.0 seconds
and 8.7 seconds (difference, 2.3; 95% CI, −1.7 to 5.0); and the median speeds on the
gait-speed test were 1.03 m per second and 1.10 m per second (difference, 0.13; 95%
CI, −0.25 to 0.30). Placebo-mode testing confirmed that improvements were due to
treatment-mode stimulation. Among the 6 participants who were also assessed at 1
year, the median within-participant changes from baseline to 1 year were generally
consistent with results at 6 months. Implantation caused ipsilateral hearing loss, with
the air-conducted pure-tone average detection threshold at 6 months increasing by 3
to 16 dB in 5 participants and by 74 to 104 dB in 3 participants. Changes in partic-
ipant-reported disability and quality of life paralleled changes in posture and gait.
CONCLUSIONS
Six months and 1 year after unilateral implantation of a vestibular prosthesis for
bilateral vestibular hypofunction, measures of posture, gait, and quality of life
were generally in the direction of improvement from baseline, but hearing was
reduced in the ear with the implant in all but 1 participant. (Funded by the Na-
tional Institutes of Health and others; ClinicalTrials.gov number, NCT02725463.)
n engl j med 384;6  nejm.org  February 11, 2021 521
The New England Journal of Medicine
Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

A
pproximately 1.8 million adults bones, internal auditory canals, and brain. Exclu-
worldwide have severe bilateral vestibular sion criteria were causes of vestibular reflex dys-
hypofunction that results in chronic dis- function unrelated to the labyrinth such as acous-
equilibrium, oscillopsia, postural instability, and tic neuroma and cerebellar atrophy, the current
unsteady gait owing to failure of vestibular re- use of substances that suppress the vestibular
flexes that stabilize the eyes, head, and body.1 reflexes, and conditions expected to preclude safe
Because affected persons must devote conscious participation in the study.14 Eight participants
effort to walking without falling, they have cogni- underwent implantation of the device. A ninth
tive distraction, a risk of falling that is 31 times participant completed the screening process but
greater than that of an unaffected person, and the did not undergo implantation of the device and
social stigma of “walking like a drunk.”2 They was therefore excluded from the analyses. The
also have a disease-related economic burden and eight participants were the only persons who
a reduction in health-related quality of life that received these implants at our institution.
are similar to those caused by adult-onset severe
bilateral hearing loss or renal insufficiency.3 Study Oversight
Current practice involves instructing these pa-This single-center study, conducted at Johns Hop-
tients to perform vestibular rehabilitation exer- kins Hospital, was a prospective, nonrandom-
cises, to avoid the use of ototoxic medications ized, single-group study in which participants
such as aminoglycosides and sedating medica- served as their own controls. It was designed by
tions such as meclizine and benzodiazepines, the investigators with input from the sponsors,
and to avoid activities that increase the risk of in-
including Labyrinth Devices, MED-EL (which, with
jury (e.g., swimming, walking in poorly lit areas, Labyrinth Devices, jointly developed the device
and driving). Noninvasive devices that use sound, evaluated in the study), and the National Insti-
skin vibration, or galvanic stimulation4-7 can con-tutes of Health (NIH). Reviewers from the NIH,
vey enough information regarding body motion the Food and Drug Administration, the institu-
and orientation to reduce postural instability; tional review board at Johns Hopkins School of
however, they do not facilitate gait.5 Medicine, and an independent data and safety
A total of 22 patients at three other centers monitoring board also provided input on the
were reported to have undergone implantation of study design. Labyrinth Devices provided the
modified cochlear implants that use electrodes implant systems, equipment, and funding to
to electrically stimulate branches of the vestibu- cover the participants’ travel expenses. MED-EL
manufactured the implanted devices and some
lar nerve; such devices have been tested exclusively
in the laboratory setting.8-12 We conducted a studyexternal components. The first six authors, the
to measure the clinical effects of a unilateral ninth and tenth authors, and the last author col-
vestibular implant that provides long-term arti- lected the data. The first, second, and last au-
ficial sensation of head motion 24 hours per day thors analyzed the data and wrote an earlier
version of the manuscript. All the authors vouch
by electrically stimulating vestibular nerve branch-
es that innervate the semicircular canals.13 for the accuracy and completeness of the data
and analyses and for the fidelity of the study to
the protocol, which is available with the full text
Me thods
of this article at NEJM.org. Confidentiality agree-
Participants ments were in place between the sponsor and
Persons who had had adult-onset bilateral ves- Johns Hopkins University.
tibular hypofunction for at least 1 year and had
undergone at least 6 months of vestibular reha- Study Design
bilitation therapy were eligible for enrollment if Testing was performed in an otolaryngology
they met the following criteria: severe bilateral clinic at baseline (<1 week before surgery) and
vestibular hypofunction confirmed by clinical repeated approximately 3 weeks after unilateral
examination and objective tests of semicircular implantation of the device. The device was then
canal function; hearing in at least one ear that activated to begin electrical stimulation to each
was adequate to support communication; and of the three semicircular canal branches 24 hours
normal findings on magnetic resonance imag- per day, as described below. Testing was re-
ing and computed tomography of the temporal peated 6 months and 1 year after implantation.

522 n engl j med 384;6  nejm.org  February 11, 2021

The New England Journal of Medicine

Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 Vestibular Implant

Active stimulation occurred continuously between The power and control unit is typically worn on
visits. At the post-activation visits, we performeda lanyard (Fig. 1; and Fig. S1 in the Supplemen-
posture and gait tests both with the device set in tary Appendix, available at NEJM.org) and uses
its usual treatment mode (in which it provided a rechargeable battery that is exchanged daily.
information on the rotational velocity of the The surgery involves creating a subperiosteal
head) and during placebo-mode stimulation (when pocket and bone well to hold the stimulator and
pulse rate and amplitude were kept constant, its inductive receiver, creating a path for the
regardless of head motion). The tests were per- electrode leads of the stimulator to run through
formed during each of the two stimulation modes the temporal bone to the labyrinth, inserting the
on the same day, in random order, with partici- electrode arrays through holes (approximately
pants and assessors unaware of stimulation mode. 0.6 mm in diameter) drilled into each semicircu-
lar canal ampulla, and packing fascia and bone
Device and Implantation Surgery chips around the arrays. A reference electrode is
The device is a self-contained system that pro- placed either in the labyrinth common crus or in
vides artificial sensation of head rotation by a subperiosteal pocket. All the operations were
electrically stimulating the three semicircular performed by one surgeon (the last author).
canal branches of the vestibular nerve (Fig. 1).
The design of the device, the surgical technique, Outcomes
and the stimulation protocols and adjustments There were five outcomes related to posture and
have been described previously.13 Similar to a gait; four participant-reported outcomes related
cochlear implant system, the device comprises to dizziness, disability, and quality of life; and
an implanted stimulator, an electrode array, and four outcomes related to hearing. Published mini-
an external processor. The stimulator is implant- mally important differences between two mea-
ed unilaterally behind the ear that has poorer surements at different times (defined as half the
baseline hearing and vestibular function. The pro- standard deviation for normative data15) were used
cedure is performed in a single operation while to provide context for the interpretation of indi-
the patient is under general anesthesia. Electri- vidual and median within-participant changes.
cal pulses are delivered to the semicircular canal
branches of the vestibular nerve through three Posture and Gait
stimulating electrodes (chosen from a total of The balance subtest (Balance Subtest 5) of the
nine electrodes implanted in the three canals) Bruininks–Oseretsky Test of Motor Proficiency,
and one reference electrode. The rotational veloc- Second Edition, comprises seven standing tasks
ity of the head is encoded in three dimensions (standing on the floor on one foot and on both
by the modulation of pulse rates and amplitudes feet and on a low balance beam on one foot,
of stimulation through the electrodes. At the time each with eyes open and with eyes closed; and
of device activation and at subsequent testing standing heel to toe on a low balance beam with
sessions, stimulation settings are adjusted as eyes open) and two walking tasks (walking for-
needed to maximize strength and selectivity of ward on the floor, with and without touching
vestibular nerve–branch stimulation, as indicated heel to toe.16 For each task, the better of two
by the speed and direction of the perceived head attempts is scored on a scale of 0 to 4 on the
motion and reflex head and eye movements. basis of time (standing tasks) or step count
The external processor includes two compo- (walking tasks) until failure (i.e., moving from
nents: a head-worn unit and a power and control the set position, opening the eyes when they
unit. The head-worn unit, which is held on the should be closed, or stepping off the line). The
scalp by three magnets, senses angular and lin- sum of the scores across the nine tasks ranges
ear head motion (only the former is used for the from 0 to 36, with higher scores indicating bet-
prosthesis) with a three-axis motion sensor and ter performance. The minimally important dif-
transmits power and commands wirelessly across ference is 1.55 points for young adults.16 We
the scalp to the implanted stimulator. The power found no published normative data for adults
and control unit sends data to a computer to older than 21 years of age.
allow for adjusting the device settings, controls The modified Romberg test involves having
and provides power to the unit worn on the the participant stand for as long as possible on
head, and alerts the user when its battery is low. a foam pad with feet together, arms crossed,

n engl j med 384;6  nejm.org  February 11, 2021 523

The New England Journal of Medicine
Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

A The Multichannel Vestibular Implant System

Implanted vestibular stimulator device

External processor
Magnets for
external processor

Head-worn unit
Inductive link coil
with three-axis
motion sensor
(gyroscope)
Electrode array
in inner ear
Stimulator

Lead (through temporal bone)

Electrodes Lanyard
Reference
Power and control unit
Posterior
Semicircular canals
Posterior
Anterior
Crista ampullaris in
Anterior ampulla of semicircular canal
and horizontal
Cupula
Vestibular nerve Horizontal
Damaged hair cells

Cochlear nerve
Electrodes
ENDOLYMPH
Anterior
and horizontal
Reference
Vestibular nerve
VIEW Anterior
branch
Electrical
current Electrodes
Posterior Nerve
stimulation

VIEW Posterior
Horizontal

B Axes of Head Rotation and Corresponding Canal Stimulation

Normal three-axis inertial sensation

Head-worn
unit Anterior Posterior Horizontal
Electrical semicircular canal semicircular canal semicircular canal
Device stimulation of
movement
corresponding
crista ampullaris

Three-axis motion
sensor (gyroscope)

524 n engl j med 384;6 nejm.org February 11, 2021

The New England Journal of Medicine

Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 Vestibular Implant

Figure 1 (facing page). Components and Mechanism velocity in meters per second; a faster speed in-
of a Vestibular Implant. dicates better performance. In a study involving
To bypass damaged hair cells in dysfunctional semicir- 558 older adults, the mean (±SD) gait speed was
cular canals, a vestibular implant electrically stimulates 1.00±0.23 m per second.24 The minimally impor-
vestibular nerve branches with the use of electrical cur- tant difference is 0.12 m per second.24
rent pulses that vary in rate and amplitude depending
on head rotation speed and axis. The multichannel ves-
tibular implant system (Panel A) comprises an implanted Participant-Reported Outcomes
stimulator with electrodes designed for insertion in the The Dizziness Handicap Inventory determines
semicircular canals; an external head-worn unit, which how often respondents have dizziness-related dif-
senses head rotation and transmits power and command ficulty or distress in 25 scenarios, each on a scale
signals to the stimulator with the use of a transcutane-
that ranges from 0 (no distress) to 4.25 The over-
ous inductive link; and a power and control unit that
houses a battery and a microprocessor that stores stim- all score is the sum of all the responses; scores
ulus settings and controls pulse timing. To mimic nerve range from 0 (no respondent-perceived dizziness
activity patterns that normally represent the speed and handicap) to 100. On the Vestibular Disorders
three-dimensional axis of head rotation, the system sep- Activities of Daily Living questionnaire, respon-
arates head rotational velocity into three components,
dents rate their perceived disability for 28 activi-
each aligned with the axis of one semicircular canal, and
then stimulates each canal separately (Panel B). ties on a scale of 1 (“independent”) to 10 (“too
difficult; no longer perform”). The overall score
is the median response, which ranges from 1
(least disability) to 10 (most disability).26,27 The
and eyes closed. The time to failure (i.e., moving minimally important difference for the Dizziness
out of position or opening the eyes) is measured Handicap Inventory is 18 points; a minimally
for a maximum of 30 seconds for each of two important difference is not specified for the
attempts, and the longer duration of the two is Vestibular Disorders Activities of Daily Living
reported.17 The minimally important difference questionnaire.28
is 4.21 seconds.18 The 36-Item Short-Form Health Survey (SF-36)
The Dynamic Gait Index assesses walking on assesses general health-related quality of life on
a level surface (with the participant’s usual gait the basis of responses to 36 questions.29 Its scor-
and while changing gait speed, turning the head ing protocol does not specify a composite score.
left and right, moving the head up and down, However, SF-36 data can be transformed to a
stepping over obstacles, stepping around obsta- health utility index score (SF-36 utility; scores
cles, or making an abrupt 180-degree pivot) and range from 0.3 [worst] to 1.0 [best]).30-32 The
walking up and down stairs.19 Each of the eight Health Utilities Index Mark 3 is a 15-question
conditions is scored with the use of a standard- survey that is intended to measure health utility
ized manual on a scale of 0 to 3; the sum of the on a scale of 0 (death) to 1 (perfect health).33 The
scores ranges from 0 to 24, with higher scores Health Utilities Index Mark 3 applies greater
indicating better gait performance. The mini- weight to hearing than does the SF-36 utility, but
mally important difference is 0.75 points.20 both yield similar scores for changes in health
In the Timed Up and Go test, a seated par- utility. The minimally important difference for
ticipant is instructed to rise, walk 3 m, turn each index is 0.03 points.33,34 To obtain aspects
around, and return and sit down.21 The mean of the participant experience that were insuffi-
completion time (in seconds) for three attempts ciently described by participant-reported outcome
is recorded, with shorter durations indicating questionnaires (none of which are specific to
better performance. In a study involving 413 bilateral vestibular hypofunction), we recorded
women 65 to 85 years of age, the completion structured interviews.
time on the Timed Up and Go test was less than
12 seconds for 92% of the women and less than Audiometry
20 seconds for all the women.22 The minimally We measured air- and bone-conducted pure-tone
important difference is 1.3 seconds.23 detection thresholds and open-set word discrimi-
Gait speed is measured during four attempts nation scores (the percentage correct among sets
of walking 10 m in a well-lit, straight corridor of 50 consonant–nucleus–consonant monosyllabic
at a comfortable pace. The time to traverse the words35). The words were presented to the par-
middle 4 m is measured and reported as the mean ticipant through earphones while the participant

n engl j med 384;6  nejm.org  February 11, 2021 525

The New England Journal of Medicine
Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

was in a sound-isolation booth. A masking noise Statistical Analysis

was played to the contralateral ear at 40 dB The main results of each outcome are presented
above its bone-conduction threshold if that as medians and 95% confidence intervals of the
threshold was at least 10 dB better than that of within-participant change from preoperative base-
the testing ear. We computed pure-tone average line to 6 months and 1 year after implantation,
detection thresholds at 500, 1000, 2000, and during treatment-mode stimulation.38 We also
4000 Hz. used descriptive statistics to summarize the re-
To simulate real-world conditions, we mea- sults obtained during placebo-mode stimulation
sured sentence recognition scores on the AzBio at 6 months and 1 year after implantation. Miss-
sentences-in-noise test (scores range from 0 to ing data after implantation were imputed to the
100%, with higher scores indicating better per- median of that participant’s previous postim-
formance).36 During this speech perception test, plantation results; no preimplantation data were
a prerecorded list of 20 sentences, chosen at ran- missing or imputed. Because there was no pre-
dom without replacement from a set of 15 such specified plan to adjust confidence intervals for
lists, is presented through a speaker in front of multiple comparisons, the end points are pre-
the participant at a moderate volume (a sound sented only as point estimates with unadjusted
pressure level of 60 dB), along with simultaneous confidence intervals.
noise (comprising the sum of ten other sen-
tences) at a sound pressure level of 55 dB. We R e sult s
performed ocular and cervical vestibular-evoked
myogenic potential testing to assess the func- Participant Characteristics and Stimulation
tion of the utricle and saccule, respectively, by de- Settings
livering air-conducted 500-Hz tone bursts through Three men and five women were enrolled in the
earphones at a sound pressure level of 125 dB.37 study and underwent unilateral implantation of the
For ocular vestibular-evoked myogenic potential device. The median age was 59 years (range, 50 to
testing, the normal range is 0 to 17 μV peak-to- 66), and all the participants had adult-onset bilat-
peak amplitude, with higher values indicating eral vestibular hypofunction that had been caused
greater utricle sensitivity to sound. For cervical by the use of an aminoglycoside, with one excep-
vestibular-evoked myogenic potential testing, the tion: Participant 7 had idiopathic hypofunction
normal range is a score of 1.3 to 3 (a dimension- (Table 1). Baseline testing occurred at a median of
less ratio of response amplitude to rectified pre- 4 years (range, 2 to 23) after the onset of symp-
stimulus electromyographic activity), with high- toms. Two participants did not undergo testing
er values indicating greater saccule sensitivity to at 1 year because of difficulty traveling as a result
sound. of the pandemic of coronavirus disease 2019.

Table 1. Participant Characteristics.

Participant No. Cause of Bilateral Vestibular Hypofunction Date of Device Date of Most Days
(Sex, Age) (Year of Symptom Onset) Implant Surgery Activation Recent Test of Use*
Ear Date
1 (M, 61 yr) Gentamicin, intravenous infusion (2012) Left Aug. 12, 2016 Sept. 7, 2016 Oct. 29, 2018 812
2 (M, 57 yr) Streptomycin, bilateral intratympanic injec- Left Nov. 4, 2016 Nov. 30, 2016 Nov. 12, 2018 738
tion (2007)
3 (F, 63 yr) Gentamicin, intravenous infusion (2015) Left Feb. 3, 2017 Feb. 23, 2017 March 27, 2019 782
4 (F, 62 yr) Gentamicin, intravenous infusion (2015) Left Dec. 15, 2017 Jan. 4, 2018 Dec. 3, 2018 354
5 (F, 50 yr) Gentamicin, bilateral intratympanic injec- Right Aug. 24, 2018 Sept. 12, 2018 Feb. 26, 2020 532
tion (2009)
6 (F, 66 yr) Gentamicin, intravenous infusion (2016) Right Aug. 31, 2018 Sept. 26, 2018 March 11, 2020 532
7 (F, 52 yr) Cause unknown (2015) Left Jan. 14, 2019 Feb. 6, 2019 July 1, 2019 145
8 (M, 55 yr) Gentamicin, intravenous infusion (1996) Right Sept. 13, 2019 Oct. 9, 2019 March 4, 2020 147

* The days of use were calculated on the basis of the date of the participant’s most recent test.

526 n engl j med 384;6  nejm.org  February 11, 2021

The New England Journal of Medicine

Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 Vestibular Implant

The median duration of the surgery was 285 within-participant difference, 5.5 points, 95%
minutes (range, 246 to 305). The reference elec- confidence interval [CI], 0 to 10.0), and the me-
trode was inserted in the labyrinth common dian time on the modified Romberg test changed
crus in seven participants and in a subperiosteal from 3.6 to 8.3 seconds (difference, 5.1; 95% CI,
pocket in Participant 8. Three participants were 1.5 to 27.6). The median score on the Dynamic
discharged the same day; the remaining five Gait Index changed from 12.5 to 22.5 (differ-
were discharged the morning after surgery. ence, 10.5 points; 95% CI, 1.5 to 12.0), the me-
Summaries of the electrical stimulation set- dian time on the Timed Up and Go test from
tings that were chosen during the outpatient pro- 11.0 to 8.7 seconds (difference, 2.3; 95% CI, −1.7
gramming visit 3 weeks after implantation and to 5.0), and the median speed on the gait-speed
at 6 months and 1 year after implantation are test from 1.03 to 1.10 m per second (difference,
provided in Tables S1, S2, and S3.13 As of the 0.13; 95% CI, −0.25 to 0.30). Among the six par-
date of each participant’s most recent test, all the ticipants who were also assessed at 1 year after
participants had used their devices in treatment implantation, the median within-participant
mode every day since activation (median dura- changes from baseline to 1 year were generally
tion of use at last assessment, 532 days; range, consistent with results at 6 months. The median
145 to 812). All the participants wore the exter- within-participant changes from baseline during
nal unit 24 hours per day for at least 6 months. placebo-mode stimulation were small, with con-
Participant 2 began taking it off at bedtime 2 years fidence intervals crossing zero (Fig. 2, Tables S7
after implantation, and Participants 5, 7, and 8 through S9, and Videos 1 through 8). Videos showing
gait and posture
began taking it off at bedtime at 6 months. The results of the assessments of participant-
tests and structured
reported dizziness, disability, and quality-of-life interviews are
Outcomes measures are shown in Table 2 and Figure 3. The available at
The following missing data from individual par- median within-participant change at 6 months NEJM.org
ticipants were imputed to the median of that was in the direction of improvement for all four
participant’s previous postimplantation results: measures related to these outcomes, and confi-
the response on one question each on the 3-week dence intervals included zero for two of the four
and 6-month postimplantation Dizziness Hand- measures. The median score on the Dizziness
icap Inventory for Participant 2; the scores on Handicap Inventory changed from 79 at baseline
the AzBio sentences-in-noise test at 6 months for to 19 at 6 months after implantation (median
Participants 1 and 2; and the results of audio- within-participant difference, 50 points; 95% CI,
metric testing at 6 months for Participant 8. The 16 to 66), and the median score on the Vestibu-
missing 1-year data for the two participants who lar Disorders Activities of Daily Living assess-
had been unable to complete testing as a result ment changed from 4 to 1 (difference, 2 points;
of the pandemic were not imputed. 95% CI, 0 to 4). The median score on the SF-36
Table 2 summarizes outcome results at base- utility changed from 0.70 to 0.82 (difference,
line, at approximately 3 weeks after implanta- 0.10 points; 95% CI, 0.06 to 0.24), and the me-
tion (but before activation of the device), and dian score on the Health Utilities Index Mark 3
during treatment-mode stimulation at 6 months changed from 0.74 to 0.72 (difference, 0.07 points;
and 1 year after implantation. Figure 2 shows 95% CI, −0.22 to 0.18). Among the six partici-
individual and median within-participant chang- pants who were also tested at 1 year after im-
es in posture and gait measurements relative to plantation, the median within-participant chang-
baseline during both treatment-mode stimula- es from baseline were similar to those observed
tion and placebo-mode simulation. During treat- at 6 months. Figure S1 shows participants using
ment-mode stimulation, the median within-par- their devices in daily life, and Videos 1 through 8
ticipant change at 6 months was positive (in the show structured interviews of all eight participants.
direction of improvement) for all five tests, and The results of the audiometric outcomes are
confidence intervals included zero for three of presented in Table 2 and Figures S2 and S3. Im-
the five tests. The median score on the balance plantation caused a bimodal distribution of
subtest of the Bruininks–Oseretsky Test of Mo- hearing loss relative to baseline in the ear con-
tor Proficiency changed from 17.5 at baseline to taining the implant; the air-conducted pure-tone
21.0 at 6 months after implantation (median average detection threshold 6 months after im-

n engl j med 384;6  nejm.org  February 11, 2021 527

The New England Journal of Medicine
Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 Table 2. Study Outcomes.*

528
Baseline, Before Activation, Treatment Mode, Treatment Mode,
1 Wk before Approximately 3 Wk af- 6 Mo after 1 Yr after Within-Participant Change from Baseline
Outcome Implantation (N = 8) ter Implantation (N = 8) Implantation (N = 8) Implantation (N = 6) in the Direction of Improvement†
Treatment Mode, 6 Mo Treatment Mode, 1 Yr
after Implantation (N = 8) ­after Implantation (N = 6)
median (95% CI)
Posture‡
Bruininks–Oseretsky Test of 17.5 (9.0 to 20.0) 13.5 (7.0 to 21.0) 21.0 (11.0 to 28.0) 25.5 (12.0 to 27.0) 5.5 (0 to 10.0) 7.0 (3.0 to 13.0)
Motor Proficiency balance
subtest score
Modified Romberg test — sec 3.6 (2.0 to 8.9) 2.5 (0.9 to 8.4) 8.3 (4.4 to 30.0) 14.5 (3.5 to 30.0) 5.1 (1.5 to 27.6) 11.0 (1.5 to 25.1)
Gait§
Dynamic Gait Index score 12.5 (9.0 to 22.5) 13.5 (9.5 to 22.0) 22.5 (21.0 to 24.0) 22.0 (21.0 to 24.0) 10.5 (1.5 to 12.0) 9.5 (−0.5 to 11.0)
Timed Up and Go test — sec 11.0 (8.2 to 19.5) 11.4 (7.5 to 19.3) 8.7 (6.8 to 14.5) 9.3 (7.3 to 14.0) 2.3 (−1.7 to 5.0) 2.1 (0 to 5.5)
Gait speed — m/sec 1.03 (0.55 to 1.27) 1.08 (0.56 to 1.37) 1.10 (0.81 to 1.35) 1.22 (0.85 to 1.27) 0.13 (−0.25 to 0.30) 0.13 (−0.08 to 0.30)
The

Dizziness and disability¶

Dizziness Handicap Inventory 79 (32 to 92) 79 (40 to 94) 19 (6 to 76) 26 (0 to 40) 50 (16 to 66) 49 (32 to 70)
score
Vestibular Disorders Activities of 4 (1 to 6) 4 (1 to 6) 1 (1 to 6) 1 (1 to 6) 2 (0 to 4) 2 (0 to 4)
Daily Living assessment score
Quality of life‖
SF-36 utility score 0.70 (0.56 to 0.83) 0.71 (0.53 to 0.85) 0.82 (0.68 to 0.90) 0.84 (0.72 to 0.90) 0.10 (0.06 to 0.24) 0.11 (0.05 to 0.28)
Health Utilities Index Mark 3 score 0.74 (0.42 to 0.85) 0.47 (−0.04 to 0.83) 0.72 (0.26 to 0.97) 0.74 (0.35 to 0.97) 0.07 (−0.22 to 0.18) 0.11 (−0.45 to 0.18)
Hearing**
Pure-tone average, air-conducted 13.1 (9.0 to 47.5) 53.8 (22.5 to 114.0) 46.9 (17.5 to 112.5) 35.0 (14.0 to 85.0) −14.4 (−100.0 to −2.5) −8.1 (−72.5 to −1.3)
— dB
n e w e ng l a n d j o u r na l

Pure-tone average, bone-­ 11.3 (7.5 to 35.0) 38.1 (16.0 to 67.5) 36.9 (10.0 to 66.3) 47.5 (5.0 to 69.0) −15.0 (−57.5 to 1.3) −11.9 (−56.3 to 2.5)

The New England Journal of Medicine

of

conducted — dB
CNC word discrimination test — % 96 (78 to 100) 80 (0 to 100) 81 (0 to 96) 84 (0 to 100) −9 (−100 to −8) −5 (−96 to 12)
AzBio sentences-in-noise test — % 92 (90 to 99) 94 (68 to 100) 88 (80 to 100) 96 (82 to 100) −3 (−18 to 6) 2 (−16 to 10)

n engl j med 384;6  nejm.org  February 11, 2021

* The widths of the confidence intervals (CI) have not been adjusted for multiple comparisons and cannot be used to infer a treatment effect.

Copyright © 2021 Massachusetts Medical Society. All rights reserved.

† Median within-participant changes in the direction of improvement relative to baseline (before implantation) are shown. The sign of within-participant change relative to baseline was inverted
m e dic i n e

for the following outcomes, for which lower values indicate better performance: the Timed Up and Go test, the Dizziness Handicap Inventory, the Vestibular Disorders Activities of Daily Living
assessment, the pure-tone average conducted through air, and the pure-tone average conducted through bone. The median within-participant difference is not necessarily equal to the differ-
ence between the baseline and postimplantation median scores.
‡ Total scores on the balance subtest of the Bruininks–Oseretsky Test of Motor Proficiency range from 0 to 36, with higher scores indicating better performance.16 Time to failure on the modified
Romberg test ranges from 0 to 30 seconds, with longer durations indicating better performance.17
§ Scores on the Dynamic Gait Index range from 0 to 24, with higher scores indicating better gait performance.19 For the Timed Up and Go test, shorter durations indicate better performance.21

Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
For gait speed, a faster speed indicates better performance.
¶ Scores on the Dizziness Handicap Inventory range from 0 to 100, with lower scores indicating a lower respondent-perceived dizziness handicap.25 Scores on the Vestibular Disorders Activities
of Daily Living assessment range from 1 to 10, with lower scores indicating lower respondent-perceived disability.27
‖ Scores on the health utility index derived from data from the 36-Item Short-Form Health Survey (SF-36 utility) range from 0.3 (worst) to 1 (best).29 Scores on the Health Utilities Index Mark 3
range from 0 (death) to 1 (perfect health).33
** Air- and bone-conducted pure-tone detection thresholds were measured in the ear containing the implant; lower threshold levels indicate better hearing. Scores on the consonant–nucleus–
consonant (CNC) word discrimination test range from 0 to 100%, with higher scores indicating better performance.35 Sentence recognition scores for sound field presentation to both ears with
added noise on the AzBio sentences-in-noise test range from 0 to 100%, with higher scores indicating better performance.36
 Vestibular Implant

Participant No.: 1 2 3 4 5 6 7 8 Median change relative Minimally important

to baseline value difference

A Posture Tests
Treatment Mode Placebo Mode
15 15
Change in Bruininks–
Oseretsky Test Score

10 10

5 5

0 0

−5 −5
0 0.5 1.0 0 0.5 1.0
Romberg Test Performance (sec)

30 30
Change in Modified

20 20

10 10

0 0

0 0.5 1.0 0 0.5 1.0

Years since Implantation

B Gait Tests
Treatment Mode Placebo Mode

10 10
Change in Dynamic
Gait Index Score

0 0

−10 −10
0 0.5 1.0 0 0.5 1.0
Change in Timed Up and Go

−5 −5
Test Performance (sec)

0 0

5 5

0 0.5 1.0 0 0.5 1.0

0.4 0.4
Change in Gait Speed

0.2 0.2

0 0
(m/sec)

−0.2 −0.2

−0.4 −0.4

−0.6 −0.6
0 0.5 1.0 0 0.5 1.0
Years since Implantation

n engl j med 384;6  nejm.org  February 11, 2021 529

The New England Journal of Medicine
Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Figure 2 (previous page). Changes in Posture and Gait.

ticipant reported a fall while walking with her
Panel A shows within-participant changes in posture tests
device unpowered. One participant fractured his
during both treatment-mode stimulation and placebo- clavicle after falling while riding a bicycle with
mode stimulation relative to baseline values measured the device in treatment mode.
before implantation. Total scores on the balance subtest
of the Bruininks–Oseretsky Test of Motor Proficiency
range from 0 to 36, with higher scores indicating better Discussion
performance.16 Time to failure on the modified Romberg
test ranges from 0 to 30 seconds, with longer durations In eight participants with adult-onset bilateral
indicating better performance.17 Panel B shows within- vestibular hypofunction, prosthetic electrical
participant changes in gait tests during both treatment- stimulation of the semicircular canal branches
mode stimulation and placebo-mode stimulation rela-
of the vestibular nerve to convey an artificial
tive to baseline values measured before implantation.
Scores on the Dynamic Gait Index range from 0 to 24, sensation of rotational head movement resulted
with higher scores indicating better gait performance.19 in changes in the direction of improvement from
For the Timed Up and Go test, shorter durations indi- baseline in two of the five measures of posture
cate better performance.21 For gait speed, a higher speed and gait and in two of the four participant-
indicates better performance. In both panels, data are
shown for all eight participants at 6 months after im-
reported outcomes related to dizziness, disability,
plantation and for six participants (all except Participants 7 and quality of life at 6 months. The results at
and 8) at 1 year after implantation. Gray shading denotes ±1 1 year after implantation were similar. Changes
minimally important difference. Brackets depict the 95% in the other three posture and gait measures and
confidence intervals of the medians, which are equiva- the other two participant-reported outcomes were
lent to the ranges. The widths of the confidence inter-
vals have not been adjusted for multiple comparisons
generally in the direction of improvement, but
and cannot be used to infer a treatment effect. The y axis the confidence intervals for the differences from
for the Timed Up and Go test is inverted so that a relative baseline included zero at 6 months. Participants
increase in performance is upward for all the outcomes. reported improvements in outcomes while using
the devices continuously, since the time of de-
plantation increased (worsened) by 74 to 104 dB vice activation, at home during activities of daily
in three participants and by 3 to 16 dB in five living; thus, such findings are not limited to
participants. Word discrimination scores 6 months laboratory testing. The direction of improvement
after implantation, assessed in the ear contain- on these tests was corroborated by participant
ing the implant, were 0% in the three partici- reports of subjective improvement in the ability
pants with greater hearing loss and within 10% to perform activities of daily life that require
of the baseline scores for the other five partici- stable posture or walking without exerting con-
pants. Sound-field AzBio sentences-in-noise rec- scious effort to avoid falling.
ognition scores, which were based on the pre- Posture and gait tests performed during
sentation of sentences to both ears, remained placebo-mode stimulation showed changes at
within 18% of the baseline score for all the 6 months and at 1 year that were not signifi-
participants, a finding that reflects intact con- cantly different from baseline measurements.
tralateral hearing. No participant had detectable Although this study was not powered to support
responses on vestibular-evoked myogenic poten- statistical inference regarding the absence of
tial testing of the ear containing the implant placebo-mode stimulation effects, these data are
before or after surgery, with the exception of consistent with an interpretation that changes in
Participant 4, who had a cervical vestibular- posture and gait performance observed during
evoked myogenic potential score of 1.02 at base- treatment-mode stimulation are not a result of
line and 1.00 after implantation. placebo effects or spontaneous recovery.
Hearing loss in the ear containing the im-
Adverse Events plant occurred in seven participants, with three
Adverse events other than hearing loss included having severe hearing loss and four having di-
tinnitus, transient imbalance, dysgeusia, facial minished hearing but retaining useful hearing
twitch, and tingling that stopped when the stimu- in the ear with the implant. One factor that may
lus current was reduced (Table S10). One par- have contributed to the hearing loss is hypersen-
ticipant reported two falls after sudden discon- sitivity to bone-conducted sound, which can re-
tinuations of stimulation related to a software sult from the presence of superior canal dehis-
error that was subsequently corrected. One par- cence or other conditions in which holes in the

530 n engl j med 384;6  nejm.org  February 11, 2021

The New England Journal of Medicine

Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 Vestibular Implant

Figure 3. Changes in Dizziness, Disability, and Quality Participant No.: 1 2 3 4 5 6 7 8

of Life.
Median change relative Minimally important
Panel A shows within-participant changes in assess- to baseline value difference
ments of dizziness and disability relative to preimplan-
tation scores. Scores on the Dizziness Handicap Inven- A Dizziness and Disability
tory range from 0 to 100, with lower scores indicating

Handicap Inventory Score

a lower respondent-perceived dizziness handicap.25 0

Change in Dizziness
Scores on the Vestibular Disorders Activities of Daily
Living assessment range from 1 to 10, with lower scores −20
indicating lower respondent-perceived disability.27 Pan-
el B shows within-participant changes in quality-of-life −40
assessments relative to preimplantation scores. Scores
on the health utility index derived from data from the −60
36-Item Short-Form Health Survey (SF-36 utility) range
from 0.3 (worst) to 1 (best).29 Scores on the Health 0 0.5 1.0
Utilities Index Mark 3 range from 0 (death) to 1 (per-
fect health).33 Data are shown individually for each of 0

Change in Vestibular

of Daily Living Score

Disorders Activities
the eight participants at 6 months after implantation
and for all but two participants (those who most re- −1
cently underwent implantation) at 1 year. Gray shading −2
denotes changes within ±1 minimally important differ-
ence as defined by published standards for each test. −3
Brackets depict the 95% confidence intervals of the
−4
medians, which are equivalent to the ranges. The widths
of the confidence intervals have not been adjusted for 0 0.5 1.0
multiple comparisons and cannot be used to infer a
Years since Implantation
treatment effect.
B Quality of Life

bone walls of the labyrinth alter the fluid dy-

Change in SF-36 Utility

0.3
namics of the inner ear.39,40 Surgically opening
the ampullae to insert vestibular-implant elec- 0.2
Score

trodes may increase cochlear sensitivity to intra-

0.1
operative drilling noise. Preserving a thin bone
layer over each ampulla until all drilling is done 0.0
and then opening each without a drill may re-
duce the risk of intraoperative noise-induced 0 0.5 1.0
hearing loss. 0.2
Change in Health Utilities

In eight participants with severe bilateral ves-

Index Mark 3 Score

tibular hypofunction who received artificial sen- 0.0

sation of head movement through prosthetic

vestibular stimulation for more than 6 months, −0.2

some measures of posture, gait, and participant-

reported quality-of-life outcomes improved from −0.4

baseline and others were generally in the direc- 0 0.5 1.0

tion of improvement. Surgery for the implanta- Years since Implantation
tion resulted in reduced hearing in the ear with
the implant in seven participants.
Disclosure forms provided by the authors are available with
Supported by the National Institute on Deafness and Other the full text of this article at NEJM.org.
Communication Disorders of the National Institutes of Health A data sharing statement provided by the authors is available
(grants R01DC013536 and 2T32DC000023), Labyrinth Devices, with the full text of this article at NEJM.org.
and MED-EL. We thank the eight courageous study participants for their
Dr. Della Santina and Johns Hopkins University report hold- time, patience, and pioneering spirit; the engineers involved in
ing royalty interests in pending and awarded patents related to the development of vestibular implant systems at MED-EL; Na-
forms of technology discussed in this article, as well as equity tan S. Davidovics, Ph.D., Gene Y. Fridman, Ph.D., Dale Roberts,
interest in Labyrinth Devices, of which Dr. Della Santina is the M.S., and the other trainees, faculty, and staff at Johns Hopkins
founder and chief executive officer. Terms of this arrangement School of Medicine who facilitated this study; and the members
are managed in accordance with Johns Hopkins University poli- of the data and safety monitoring board (David J. Eisenman, M.D.,
cies on conflicts of interest. Frank R. Lin, M.D., Ph.D., and C. Matthew Stewart, M.D., Ph.D.).

n engl j med 384;6  nejm.org  February 11, 2021 531

The New England Journal of Medicine
Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 Vestibular Implant

Appendix
The authors’ full names and academic degrees are as follows: Margaret R. Chow, B.S., Andrianna I. Ayiotis, B.S., Desi P. Schoo, M.D.,
Yoav Gimmon, P.T., Ph.D., Kelly E. Lane, R.V.T., Brian J. Morris, B.S., Mehdi A. Rahman, M.S., Nicolas S. Valentin, M.S., Peter J.
Boutros, Ph.D., Stephen P. Bowditch, Au.D., Bryan K. Ward, M.D., Daniel Q. Sun, M.D., Carolina Treviño Guajardo, M.D., Michael C.
Schubert, P.T., Ph.D., John P. Carey, M.D., and Charles C. Della Santina, M.D., Ph.D.
The authors’ affiliations are as follows: the Departments of Otolaryngology–Head and Neck Surgery (M.R.C., A.I.A., D.P.S., Y.G.,
K.E.L., B.J.M., P.J.B., S.P.B., B.K.W., D.Q.S., C.T.G., M.C.S., J.P.C., C.C.D.S.) and Biomedical Engineering (M.R.C., A.I.A., B.J.M.,
P.J.B., C.C.D.S.), Johns Hopkins University School of Medicine, and Labyrinth Devices (M.A.R., N.S.V., C.C.D.S.) — both in Baltimore.

References
1. Living without a balancing mecha- Bilateral vestibulopathy: diagnostic crite- Daily Living Scale. Arch Otolaryngol Head
nism. N Engl J Med 1952;​246:​458-60. ria consensus document of the classifi- Neck Surg 2000;​126:​881-7.
2. Ward BK, Agrawal Y, Hoffman HJ, cation committee of the Bárány Society. 28. Cohen HS, Kimball KT, Adams AS.
Carey JP, Della Santina CC. Prevalence and J Vestib Res 2017;​27:​177-89. Application of the Vestibular Disorders
impact of bilateral vestibular hypofunc- 15. Wright A, Hannon J, Hegedus EJ, Ka- Activities of Daily Living Scale. Laryngo-
tion: results from the 2008 US National vchak AE. Clinimetrics corner: a closer scope 2000;​110:​1204-9.
Health Interview Survey. JAMA Otolaryn- look at the minimal clinically important 29. Jenkinson C, Coulter A, Wright L.
gol Head Neck Surg 2013;​139:​803-10. difference (MCID). J Man Manip Ther 2012;​ Short form 36 (SF36) health survey ques-
3. Sun DQ, Ward BK, Semenov YR, Carey 20:​160-6. tionnaire: normative data for adults of
JP, Della Santina CC. Bilateral vestibular 16. Bruininks RH, Bruininks BD. Bruininks- working age. BMJ 1993;​306:​1437-40.
deficiency: quality of life and economic Oseretsky test of motor proficiency, sec- 30. Lins L, Carvalho FM. SF-36 total score
implications. JAMA Otolaryngol Head ond edition. London:​Pearson, 2005. as a single measure of health-related
Neck Surg 2014;​140:​527-34. 17. Agrawal Y, Carey JP, Hoffman HJ, quality of life: scoping review. SAGE Open
4. Honegger F, Hillebrandt IM, van den Sklare DA, Schubert MC. The modified Med 2016;​4:​2050312116671725.
Elzen NGA, Tang K-S, Allum JH. The ef- Romberg Balance Test: normative data 31. Brazier J, Roberts J, Deverill M. The
fect of prosthetic feedback on the strate- in U.S. adults. Otol Neurotol 2011;​ 32:​ estimation of a preference-based measure
gies and synergies used by vestibular loss 1309-11. of health from the SF-36. J Health Econ
subjects to control stance. J Neuroeng 18. Vereeck L, Wuyts F, Truijen S, Van de 2002;​21:​271-92.
Rehabil 2013;​10:​115. Heyning P. Clinical assessment of bal- 32. Ara R, Brazier J. Predicting the short
5. Sienko KH, Whitney SL, Carender WJ, ance: normative data, and gender and age form-6D preference-based index using the
Wall C. The role of sensory augmentation effects. Int J Audiol 2008;​47:​67-75. eight mean short form-36 health dimen-
for people with vestibular deficits: real- 19. Shumaway-Cook A, Woollacott M. sion scores: estimating preference-based
time balance aid and/or rehabilitation de- Motor control:​theory and practical appli- health-related utilities when patient level
vice? J Vestib Res 2017;​27:​63-76. cations. Baltimore:​Williams & Wilkins, data are not available. Value Health 2009;​
6. Fitzpatrick RC, Day BL. Probing the 1995. 12:​346-53.
human vestibular system with galvanic 20. Herman T, Inbar-Borovsky N, Brozgol 33. Horsman J, Furlong W, Feeny D, Tor-
stimulation. J Appl Physiol (1985) 2004;​ M, Giladi N, Hausdorff JM. The Dynamic rance G. The Health Utilities Index (HUI):
96:​2301-16. Gait Index in healthy older adults: the role concepts, measurement properties and
7. Barros CGC, Bittar RSM, Danilov Y. of stair climbing, fear of falling and gen- applications. Health Qual Life Outcomes
Effects of electrotactile vestibular substi- der. Gait Posture 2009;​29:​237-41. 2003;​1:​54.
tution on rehabilitation of patients with 21. Centers for Disease Control and Pre- 34. Fryback DG, Dunham NC, Palta M,
bilateral vestibular loss. Neurosci Lett vention. Timed Up & Go (TUG). 2017 et al. US norms for six generic health-
2010;​476:​123-6. (https://www​.­cdc​.­gov/​­steadi/​­pdf/​­TUG_Test​ related quality-of-life indexes from the
8. Sluydts M, Curthoys I, Vanspauwen R, -­print​.­pdf). National Health Measurement study. Med
et al. Electrical vestibular stimulation 22. Bischoff HA, Stähelin HB, Monsch AU, Care 2007;​45:​1162-70.
in humans: a narrative review. Audiol et al. Identifying a cut-off point for nor- 35. Peterson GE, Lehiste I. Revised CNC
Neurootol 2020;​25:​6-24. mal mobility: a comparison of the timed lists for auditory tests. J Speech Hear Dis-
9. Merfeld DM, Lewis RF. Replacing ‘up and go’ test in community-dwelling ord 1962;​27:​62-70.
semicircular canal function with a ves- and institutionalised elderly women. Age 36. Spahr AJ, Dorman MF, Litvak LM, et al.
tibular implant. Curr Opin Otolaryngol Ageing 2003;​32:​315-20. Development and validation of the AzBio
Head Neck Surg 2012;​20:​386-92. 23. Ibrahim A, Singh DKA, Shahar S. sentence lists. Ear Hear 2012;​33:​112-7.
10. Rubinstein JT, Ling L, Nowack A, Nie ‘Timed Up and Go’ test: age, gender and 37. Nguyen KD, Welgampola MS, Carey
K, Phillips JO. Results from a second- cognitive impairment stratified norma- JP. Test-retest reliability and age-related
generation vestibular implant in human tive values of older adults. PLoS One 2017;​ characteristics of the ocular and cervical
subjects: diagnosis may impact electrical 12(10):​e0185641. vestibular evoked myogenic potential tests.
sensitivity of vestibular afferents. Otol 24. Brach JS, Perera S, Studenski S, New- Otol Neurotol 2010;​31:​793-802.
Neurotol 2020;​41:​68-77. man AB. The reliability and validity of 38. Campbell MJ, Gardner MJ. Calculat-
11. Guyot J-P, Perez Fornos A. Milestones measures of gait variability in community- ing confidence intervals for some non-
in the development of a vestibular im- dwelling older adults. Arch Phys Med Re- parametric analyses. Br Med J (Clin Res
plant. Curr Opin Neurol 2019;​32:​145-53. habil 2008;​89:​2293-6. Ed) 1988;​296:​1454-6.
12. Guinand N, van de Berg R, Cavuscens 25. Jacobson GP, Newman CW. The devel- 39. Iversen MM, Zhu H, Zhou W, Della
S, et al. Vestibular implants: 8 years of opment of the Dizziness Handicap Inven- Santina CC, Carey JP, Rabbitt RD. Sound
experience with electrical stimulation of tory. Arch Otolaryngol Head Neck Surg abnormally stimulates the vestibular sys-
the vestibular nerve in 11 patients with 1990;​116:​424-7. tem in canal dehiscence syndrome by
bilateral vestibular loss. ORL J Otorhino- 26. Cohen HS. Use of the Vestibular Dis- generating pathological fluid-mechanical
laryngol Relat Spec 2015;​77:​227-40. orders Activities of Daily Living Scale to waves. Sci Rep 2018;​8:​10257.
13. Boutros PJ, Schoo DP, Rahman M, et al. describe functional limitations in patients 40. Ward BK, Carey JP, Minor LB. Superi-
Continuous vestibular implant stimula- with vestibular disorders. J Vestib Res 2014;​ or canal dehiscence syndrome: lessons
tion partially restores eye-stabilizing re- 24:​33-8. from the first 20 years. Front Neurol 2017;​
flexes. JCI Insight 2019;​4(22):​e128397. 27. Cohen HS, Kimball KT. Development 8:​177.
14. Strupp M, Kim J-S, Murofushi T, et al. of the Vestibular Disorders Activities of Copyright © 2021 Massachusetts Medical Society.

532 n engl j med 384;6  nejm.org  February 11, 2021

The New England Journal of Medicine

Downloaded from nejm.org on November 21, 2022. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.