Instrumented Gait Analysis to Identify Persistent Deficits in Gait Stability in Adults With Chronic Vestibular Loss

Key Points

Question  Is chronic peripheral vestibular hypofunction associated with enduring gait deficits?

Findings  In this cross-sectional study, 13 adults with chronic vestibular loss had mean stride lengths that were 0.17 m shorter and mean peak whole-body turning velocity that was 50.4°/s slower when compared with 17 healthy adults. Instrumented gait analysis detected deficits even when observational testing suggested low fall risk.

Meaning  The results of this study suggest that adults with chronic vestibular loss have clinically and functionally meaningful gait deficits.

Importance  Regaining the ability to walk safely is a high priority for adults with vestibular loss. Thus, practitioners need comprehensive knowledge of vestibulopathic gait to design, provide, and/or interpret outcomes of interventions. To date, few studies have characterized the effects of vestibular loss on gait.

Objectives  To investigate the use of an instrumented 2-minute walk test in adults with vestibular loss, to further characterize vestibulopathic gait, and to assess whether those with chronic vestibular loss have enduring gait deficits.

Design, Setting, and Participants  This cross-sectional study, conducted between April 3, 2018, and June 27, 2019, recruited adults 20 to 79 years of age from an academic, tertiary, hospital-based, ambulatory care setting who were healthy or had confirmed unilateral or bilateral vestibular hypofunction. Of the 43 adults who were screened from convenience and referred samples, 2 declined, and 7 were excluded because of health conditions.

Exposures  The main exposure was the instrumented 2-minute walk test, which was conducted with participants using wearable inertial measurement units while they walked a 10-m path at their self-selected speed and turned 180° in their self-selected direction at either end.

Main Outcomes and Measures  The primary measures were spatiotemporal gait metrics (eg, stride length [SL] and peak whole-body turning velocity). Multivariate analysis of variance was used to assess between-group differences. Validity was assessed using the area under the curve from receiver operator characteristic analyses.

Results  Data from 17 healthy adults (mean [SD] age, 39.27 [11.20] years; 13 [76%] female) and 13 adults with vestibular loss (mean [SD] age, 60.50 [10.81] years; 6 [46%] female) were analyzed. Very large between-group differences were found for SL (left) (estimated marginal mean [SE] for healthy vs vestibular groups, 1.47 [0.04] m vs 1.31 [0.04] m; Cohen d, 1.35; 95% CI, 0.18-2.52), SL (right) (estimated marginal mean [SE] for healthy vs vestibular groups, 1.46 [0.04] m vs 1.29 [0.04] m; Cohen d, 1.44; 95% CI, 0.25-2.62), and peak turn velocity (estimated marginal mean [SE] for healthy vs vestibular groups, 240.17 [12.78]°/s vs 189.74 [14.70]°/s; Cohen d, 1.23; 95% CI, 0.07-2.40). The area under the curve was 0.79 (95% CI, 0.62-0.95) for SL (left), 0.81 (95% CI, 0.64-0.97) for SL (right), and 0.86 (95% CI, 0.72-0.99) for peak turn velocity.

Conclusions and Relevance  In this cross-sectional study, instrumented gait analysis had good discriminative validity and revealed persistent deficits in gait stability in those with chronic vestibular loss. The findings of this study suggest that these clinically and functionally meaningful deficits could be targets for vestibular rehabilitation.

Vestibular disorders affect 5% to 10% of the general population^1,2 and cause functional impairments in vision,³ balance,⁴ and gait.⁵ Although the effects of peripheral vestibular loss on balance⁶ and gaze stability^7,8 are well documented, less attention has been devoted to its effect on gait.^9-11 Greater understanding of vestibulopathic gait is needed because regaining the ability to walk safely is a high priority for patients.¹²

The functional effect of vestibular loss on walking is typically evaluated using observational assessments.^12-17 Instrumented gait analysis in persons with vestibular loss is possible using pressure switches,^5,18 3-dimensional motion capture systems,^19-21 pressure mats,^22,23 gyroscopes,²⁴ and inertial measurement units (IMUs).^25-28 Although many instrumented methods are not available in most clinical settings, wearable technologies²⁹ are becoming more accessible.

The extent of vestibular loss, temporal aspects of recovery, and the role of central compensation^12,30 are important considerations for interpreting vestibulopathic gait deficits. Abnormalities in the timing of spatiotemporal gait parameters^18,24 and reduced gait velocity (GV)³¹ have been observed in the short term. Patients with subacute unilateral vestibular loss (UVL) who have undergone vestibular schwannoma resection have reduced peak head rotation amplitude and velocity as well as impaired head-trunk coordination while walking at their self-selected speed.^27,28 Although some studies^18,31 of adults with subacute UVL suggest that spatiotemporal gait parameters normalize during self-paced, linear gait in the light, deficits in fast walking and walking in the dark persist for at least 3 months after vestibular neurectomy.³¹ Furthermore, adults with chronic (>3 months) UVL and bilateral vestibular loss (BVL) have a clinically meaningful reduction in self-paced, linear walking speed of 0.22 to 0.25 m/s^21,22 as well as imbalance during head turns and difficulty negotiating obstacles.²²

Many patients^32,33 are not referred for vestibular rehabilitation (VR) until the long-term recovery phase; thus, further characterization of vestibulopathic gait in patients more than 3 months after onset is needed. In addition, because the vestibular system influences gait by sensing linear and angular movement,⁹ specific attention to turning metrics is warranted.

The objectives of this study were to investigate the use of an instrumented 2-minute walk test (i2MWT) in adults with chronic vestibular loss, characterize spatiotemporal aspects of vestibulopathic gait, and compare the ability to detect persistent gait abnormalities using observational and IMU-based gait analysis. We hypothesized that GV and turning dynamics would discriminate adults with vestibular loss from healthy adults, that those with BVL would perform worse compared with those with UVL, and that IMU-based gait analysis would detect gait deficits even when observational testing does not.

This cross-sectional study was completed in an academic, tertiary, ambulatory care setting. Data were collected between April 3, 2018, and June 27, 2019, by 1 of the investigators (C.R.G.); all data were deidentified for the other authors. All participants provided written informed consent. All procedures were conducted in accordance with the provisions of the Declaration of Helsinki.³⁴ The study was approved by the University of Wisconsin–Madison Health Sciences Institutional Review Board. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

We recruited adults who were 20 to 79 years of age; could speak English fluently; were able to function independently; had no neurologic, musculoskeletal, visual, or pain conditions that would preclude accurate testing; were able to stand for 20 minutes without sitting and walk 6.1 m without assistance; could withhold antivertigo, sedative, and narcotic or barbiturate medications and abstain from drinking alcohol for 48 hours before the study visit; and would not be pregnant while participating. All participants were required to have at least 75% of active range of motion in the lower extremities, normal bilateral lower-extremity strength,³⁵ and normal bilateral lower-extremity sensation.³⁶ Healthy participants without vestibular loss were recruited from the community. Adults with vestibular loss were recruited from outpatient clinics and were enrolled if they had a confirmed diagnosis of UVL or BVL (eAppendix 1 in the Supplement). All patients underwent a clinical examination, the noninstrumented dynamic visual acuity test (DVAT),³ the Sensory Organization Test (SOT),³⁷ and the measures that follow (eAppendix 2 in the Supplement). Each participant received a stipend.

Participants completed the Activities-Specific Balance Confidence Scale (ABCS),³⁸ Dizziness Handicap Inventory (DHI),³⁹ Visual Vertigo Analog Scale (VVAS),⁴⁰ and Vestibular Activities and Participation Measure (VAPM)⁴¹ within 48 hours of the start of the study visit. Only the mean or total scores are reported for the ABCS, DHI, VVAS, and VAPM. Healthy participants did not complete the disease-specific VAPM.

The Functional Gait Assessment (FGA)⁴² consists of 10 walking tasks, each scored on an ordinal scale of 0 to 3 points, with higher scores indicating better performance or better balance. The FGA was always completed before the i2MWT. The optimal total score is 30. The total score and gait speed, which was manually recorded during the first item of the FGA (walking, eyes open), are reported.

One trial of the i2MWT²⁹ was conducted while participants wore 3 IMUs (IMU, version 1.0; IMU software, version 2.0; APDM Inc). One sensor was placed on the torso at the level of the fourth and fifth lumbar intervertebral disks using the manufacturer-provided strap. The remaining sensors were secured on the dorsum of each foot. Participants wore their preferred, flat-soled shoes and walked within a 10 × 0.4-m path at their preferred speed and turned 180° in their preferred direction at either end. The metrics of interest were GV in meters per second, cadence in steps per minute, double support as a percentage of the gait cycle, stride length (SL) in meters, turning angle in degrees, peak turn velocity (PTV) and mean turn velocity (MTV) in degrees per second, turn duration (TD) in milliseconds, and the number of steps to turn (eAppendix 2 in the Supplement). Gait cycles and the number of turns were also recorded.

All analyses were conducted post hoc using R software, version 3.5 (R Foundation for Statistical Computing).⁴³ The α level was 0.05. Between-group comparisons for age, sex, body mass index (calculated as weight in kilograms divided by height in meters squared), height in meters, and fall history were analyzed with 2-tailed t tests or χ² tests. Demographic differences between subgroups were evaluated using analysis of variance with the Tukey honestly significant difference or a χ² test. Equipment malfunction during the i2MWT led us to exclude data from 4 adults with vestibular loss.

Between-group and between-subgroup differences for the i2MWT, as well as secondary self-report and capacity-based measures, were examined with 1-way, multivariate analysis of variance controlling for age and sex. Effect sizes for specific i2MWT metrics were calculated using the Cohen d (with 0-0.2 indicating small, 0.3-0.5 indicating medium, 0.6-0.8 indicating large, and ≥0.9 indicating very large).⁴⁴ In addition, z scores (SDs from the mean of healthy adults in this study) for the i2MWT metrics were calculated.⁴⁵ The minimal detectable change for each i2MWT metric was calculated using 1.96 × √2 × SEM.⁴⁶

The discriminative ability of SL, PTV, and the FGA total score was assessed with the area under the curve from receiver operator characteristic (ROC) curve analyses. The sensitivity and specificity (95% CI) for these metrics were calculated from contingency tables created based on their optimal ROC thresholds. The sensitivity and specificity of the FGA to detect a history of falls using a cutoff score of 22 of 30 was also assessed.⁴⁷ The Pearson method was used to assess associations between PTV and self-report and other capacity-based measures.

Data from 17 healthy adults (mean [SD] age, 39.27 [11.20] years; 13 [76%] female) and 13 adults with vestibular loss (mean [SD] age, 60.50 [10.81] years; 6 [46%] female) (whose onset ranged from 3 months to 12 years before enrollment) were analyzed (Table 1). Very large between-group differences were found for age (see above) and fall history (1 [6%] for healthy adults vs 5 [39%] for adults with vestibular loss). Mean (SD) height (1.72 [0.09] m for healthy adults vs 1.75 [0.08] m for adults with vestibular loss) and body mass index (25.96 [4.36] for healthy adults vs 26.16 [4.90] for adults with vestibular loss) were similar for both groups. Between-group and between-subgroup analyses for self-report and capacity-based measures, including the i2MWT, were performed while controlling for age and sex. The width of the CIs for each reported effect size suggests uncertainty about the precision of the estimates from this preliminary study.

Very large between-group differences in estimated marginal means (SEs) were found for the ABCS (16.52 [6.07]), DHI (−30.03 [6.78]), VVAS (−21.99 [7.99]), DVAT (horizontal: −3.05 [1.07]; vertical: −3.91 [1.03]), SOT (17.1 [6.72]), and FGA (4.72 [1.82]) (Table 2). Healthy adults had better results on all self-report and capacity-based measures vs adults with vestibular loss (ABCS: 98.99 [3.52] vs 82.47 [3.90]; DHI: 0.04 [3.93] vs 30.07 [4.36]; VVAS: −1.73 [4.64] vs 20.26 [5.14]; DVAT [horizontal]: 1.28 [0.62] vs 4.33 [0.69]; DVAT [vertical]: 0.8 [0.6] vs 4.71 [0.66]; SOT: 82.82 [3.9] vs 65.73 [4.32]; FGA: 28.86 [1.05] vs 24.14 [1.17]; gait speed: 1.43 [0.07] vs 1.26 [0.08]; gait cycles: 50.55 [1.88] vs 45.69 [2.21]; and number of turns: 15.66 [0.6] vs 13.68 [0.71]).

Very large differences in estimated marginal means (SEs) were found between healthy adults and those with UVL for the DHI (−26.09 [7.62]), VVAS (−18.77 [9.1]), and vertical DVAT (−2.28 [0.95]) (Table 3). Differences in estimated marginal means (SEs) between healthy adults and those with BVL for the ABCS (24.92 [6.16]), DHI (−34.21 [7.71]), VVAS (−25.38 [9.21]), DVAT (horizontal: −5.00 [0.96]; vertical: −5.64 [0.96]), SOT (28.72 [6.19]), and FGA (8.39 [1.47]) were very large (Table 3). The differences for the ABCS (16.34 [6.16]), DVAT (horizontal: −3.81 [0.96]; vertical: −3.36 [0.96]), SOT (22.63 [6.19]), and FGA (7.16 [1.47]) between adults with UVL and those with BVL were very large (Table 3). Small to large differences were found for gait speed (effect size, 0.76; 95% CI, −0.37 to 1.89), gait cycles (effect size, 0.63; 95% CI, −0.15 to 1.42), and number of turns (effect size, 0.81; 95% CI, 0.01-1.60) (Table 2 and Table 3).

Analyzing values for GV, cadence, and SL that were normalized to height⁴⁸ did not alter the conclusions to be drawn; thus, only the analysis of the raw values for these metrics are reported (Table 4). Adults with vestibular loss had estimated marginal means (SEs) indicating slower GV (1.25 [0.07] m/s), shorter SL bilaterally (left: 1.31 [0.04] meters; right: 1.29 [0.04] meters), slower PTV (189.74 [14.17]°/s), and longer TD (1180 [89] milliseconds) compared with healthy adults; however, post hoc analysis revealed that the very large between-group difference in age influenced these results. Even after age and sex were controlled for, the effect sizes for the between-group differences for GV (0.98; 95% CI, −0.16 to 2.12), SL (left: 1.35; 95% CI, 0.18-2.52; right: 1.44; 95% CI, 0.25-2.62), TD (−0.97; 95% CI, −2.12 to 0.17]), and PTV (1.23; 95% CI, 0.07-2.40) were very large. The z scores indicate that the performance of the group with vestibular loss differed from the healthy group by more than 1 SD (eg, −1.05 for SL [left], −1.13 for SL [right], and −1.23 for PTV). Between-group differences for each i2MWT metric (eg, mean [SD], −0.64 [1.24] m for SL [left], 0.16 [0.07] m for SL [right], and 50.43 [22.03]°/s for PTV) exceeded the respective minimal detectable change values calculated from this population (eg, 0.01 m for SL [left], 0.01 m for SL [right], and 4.72°/s for PTV).

Very large estimated marginal mean (SE) differences were found between healthy adults and those with UVL for GV (0.21 [0.12] m/s), SL (0.18 [0.08] m [left] and 0.17 [0.07] [right]), and PTV 46.58 (25.31)°/s). The estimated marginal mean (SE) differences between healthy adults and those with BVL were very large for SL (0.15 (0.08) [left] and 0.17 [0.07] [right], TD (−445 [141] milliseconds), PTV (54.49 [25.62]°/s), and MTV (55.95 [20.82]°/s). Very large estimated marginal mean (SE) differences were found between adults with UVL and those with BVL for turning angle (7.88 [3.16]°), TD (−378 [132] milliseconds), and MTV (45.82 [19.61]°/s) (Table 5).

The results of the post hoc analysis of age and sex influenced the selection of i2MWT metrics to be analyzed for discriminative ability. The area under the curve was 0.79 (95% CI, 0.62-0.95; 1.46 m) for SL (left), 0.81 (95% CI, 0.64-0.97; 1.42 m) for SL (right), 0.86 (95% CI, 0.72-0.99; 229.5°/s) for PTV, and 0.89 (95% CI, 0.77-1.00; 27 of 30 points) for the FGA. The sensitivity for these metrics to detect vestibular dysfunction was 100% (95% CI, 75%-100%) for SL (left), 100% (95% CI, 75%-100%) for SL (right), 100% (95% CI, 75%-100%) for PTV, and 69% (95% CI, 39%-91%), and 100% (80%, 100%) for the FGA, and the specificity was 53% (95% CI, 29%-77%) for SL (left), 59% (95% CI, 33%-88%) for SL (right), 71% (95% CI, 44%-90%) for PTV, and 100% (95% CI, 80%-100%) for the FGA. When evaluating the ability to detect a history of falls using an FGA total score of 22 of 30,⁴⁸ the sensitivity was 38% (95% CI, 14%-68%) and the specificity was 100% (95% CI, 80%-100%).

Assessment of the concurrent validity of SL and PTV revealed large to very large effect sizes for correlations with self-report measures: ABCS mean score (SL [left]: 0.87; 95% CI, 0.08-1.67; SL [right]: 1.01; 95% CI, 0.19-1.82; PTV: 1.39; 95% CI, 0.5-2.27), DHI total score (SL [left]: −1.06; 95% CI, −1.89 to −0.24; SL [right]: −1.22; 95% CI, −2.07 to −0.37; PTV: 1.39; 95% CI, 0.5-2.27), VVAS mean score (SL [left]: −0.87; 95% CI, −1.67 to −0.08; SL [right]: −0.77; 95% CI, −1.55 to 0.01; PTV: −0.82; 95% CI, −1.61 to −0.03). Similarly, large to very large effect sizes were found for correlations of SL and PTV with capacity-based measures: DVAT horizontal and vertical (SL [left]: −0.85; 95% CI, −1.64 to −0.06; SL [right]: −0.95; 95% CI, −1.76 to −0.15), DVAT horizontal (PTV: −1.62; 95% CI, −2.56 to −0.69), DVAT vertical (PTV: −1.54; 95% CI, −2.46 to −0.62), SOT composite score (SL [left]: 1.19; 95% CI, 0.34-2.03; SL [right]: 1.39; 95% CI, 0.50-2.27; PTV: 1.62; 95% CI, 0.69-2.56), and the FGA total score (SL [left]: 1.39; 95% CI, 0.50-2.27; SL [right]: 1.58; 95% CI, 0.65-2.51; PTV: 1.85; 95% CI, 0.86-2.85) (eTable 1 in the Supplement).

One healthy adult reported visually induced dizziness on the VVAS. Performance was above the optimal thresholds from the ROC analyses for SL (left and right) for 47% of healthy participants and PTV for 70% of healthy participants (eTable 2 in the Supplement).

For self-report measures, 30% of individuals with vestibular loss scored higher than the normal cutoff score (80%) on the ABCS,³⁸ 54% scored mild disability or worse (30 of 100) on the DHI,⁴⁹ 54% had visually induced dizziness (abnormal VVAS score),⁵⁰ 24% scored mild activity limitations and participation restrictions (≤1.0) on the VAPM,⁴¹ and 38% scored higher than the falls risk cutoff score (22 of 30) on the FGA.⁴⁷ On the i2MWT, none of the adults with vestibular loss had values above the ROC thresholds for SL (left), SL (right), and PTV. Thus, all those with FGA total scores greater than 22 of 30 had abnormal SL (bilaterally) and PTV. In addition, 46% had PTV of less than 180°/s⁹ (eTable 3 in the Supplement).

In this cross-sectional study, many adults with chronic vestibular loss had persistent deficits, particularly for SL and PTV while walking at their self-selected speed. Both SL and PTV discriminated healthy adults from adults with vestibular loss; however, turn angle, TD, and MTV differentiated adults with UVL from those with BVL. In addition, instrumented gait analysis detected persistent deficits in walking, even when observational gait testing suggested the person may be at low risk of falling.

This study’s findings of persistent gait impairments during prolonged walking extend those of prior studies^19-21,31 in which forward gait was assessed in adults with vestibular loss who walked a straight path 10 m or shorter. In addition, the data regarding whole-body turning dynamics add to the small body of work in adults with vestibular loss that has documented gait deficits during yaw-plane head movements,²⁵ head-trunk incoordination while performing selected tasks from the FGA,²⁷ and reduced active head movements during community ambulation.²⁸ Furthermore, this study’s findings are consistent with those of prior studies^22,23 that documented deficits in spatiotemporal aspects of gait when adults with imbalance attributable to vestibular loss or other causes performed complex gait tasks while walking on a pressure mat.

The mean GV of adults with chronic vestibular loss in this study is consistent with that reported in prior studies.^22,51 The difference in GV between the groups in this study exceeds the range for the minimal clinically important difference for gait speed of 0.10 to 0.20 m/s⁵²; however, the older age of the adults with vestibular loss contributed to their reduced GV.

The reduction in SL and lower PTV of adults with vestibular loss suggests that vestibular loss may have a lasting effect on gait stability. Although the minimal clinically important difference has not been established for SL and PTV from the i2MWT, the observed effect sizes were very large.

Compared with adults who do not fall, those who fall have a very large reduction in SL.⁵³ In addition, SL that is normalized to height has 93% sensitivity and 53% specificity as a marker of recurrent falls in community-dwelling older adults.⁵⁴ Further research is needed to determine whether SL is a marker for falls risk in adults with vestibular loss.

The lingering deficits in the turning dynamics of adults with vestibular loss are functionally meaningful. The PTV for half of adults with chronic vestibular loss, particularly those with BVL, was far below 180°/s.⁹ Although we did not dictate which direction participants should turn, lesion localization does not appear to affect which direction adults with vestibular loss choose to turn.²⁸ This study’s data suggest that many adults with chronic vestibular loss have adapted to turn at speeds that help them avoid imbalance or to reduce the impact of oscillopsia on gait.⁵⁵ In addition, the moderate associations of PTV with balance-related confidence, visual acuity during movement, and gait-related balance suggest that turning dynamics may play a role in fall risk for adults with vestibular loss. Spatiotemporal gait metrics obtained during straight walking and turning are similarly associated with future falls.⁵⁶ Evidence from community-dwelling older adults demonstrates that prospective fallers turn less frequently, more slowly, and with less consistency in turning angle than individuals who do not fall.⁵⁷ Self-limiting turning speed may reduce the risk of falls for adults with vestibular loss⁵⁸ but may also result in insufficient vestibular stimulation to drive central compensation.²⁸

Clear associations between the gait abnormalities detected with the i2MWT and fall risk have not been determined. Half of the adults with vestibular loss who reported at least 1 fall in the prior 6 months had total scores above the only established cutoff score (22 of 30 points) for fall risk on the FGA.⁴⁷ In addition, gait abnormalities were detected with IMUs for all adults with vestibular loss who reported a fall and all adults with vestibular loss who had a total FGA score greater than 22 of 30. On the basis of this threshold, the FGA total score had poor sensitivity for detecting a history of falls in this population. Thus, the study data suggest that observational gait analysis alone may not be sufficient for documenting the potentially serious association of vestibular loss with gait.

Gait analysis with IMUs has good to excellent reliability for healthy, young adults while they are walking on a treadmill, on a split-belt treadmill, and over ground.⁵⁹ In addition, gait analysis using IMUs for the Timed Up and Go test has fair to good reliability in adults with vestibular loss.²⁶ Validity is improved when IMUs are placed on the dorsum of the foot, as was the case in this study, rather than on the lower shank.⁵⁹ The preliminary findings of good to excellent discriminative validity for SL and PTV from the i2MWT extend prior work²⁶ indicating that IMU-based gait data collected during the Timed Up and Go test discriminate adults with vestibular loss based on fall history. Furthermore, these data suggest that peak whole-body turning velocity has moderate to strong concurrent validity with self-report and capacity-based measures of balance function.

At the group level, these data are consistent with previous studies that reported reduced GV^20,22,23,31 and increased fall risk^14,15,17 for adults with vestibular loss. However, the results of the current study also agree with others who found that GV was normal for age and sex in 33% of adults with UVL and 39% of those with BVL at the start of VR.³² In addition, the current study data agree with these same authors,³² who also reported that 21% of adults with UVL and 22% of those with BVL scored above the threshold for increased fall risk based on observational gait analysis at their baseline assessment for VR.³³ Therefore, although some generalizations regarding vestibulopathic gait may be possible, practitioners should always consider the specific data gathered from each patient, particularly because customized VR programs lead to superior outcomes compared with generic exercises.⁶⁰

Central vestibular compensation is idiosyncratic⁶¹; however, adaptive changes in locomotor patterns or increased coordination of the head and trunk may contribute to the recovery of gait after vestibular loss.⁶² Before undergoing ablative procedures, persons with Meniere disease, who experience a gradual loss of vestibular function over time, appear to benefit from compensatory strategies that reduce errors in turning angle⁶³ and walking trajectory.³¹ However, gait deficits associated with turning angle,⁶³ walking trajectory,³¹ and head-trunk coordination^27,28 are present in the immediate postoperative period. The current study data indicate that significant deficits persist long after the onset of vestibular loss.

Beyond central vestibular compensation, gait deficits can be addressed through walking exercises, which are an important component of VR.¹² More specific characterization of vestibulopathic gait using IMU-based methods may provide practitioners with further insight into designing effective interventions. The current study data indicate that gait stability and turning dynamics should be targets of VR for adults with chronic vestibular loss. Preliminary evidence suggests that spatiotemporal gait parameters normalize when persons with vestibular loss are exposed to noisy galvanic vestibular stimulation⁶⁴ or complete paced walking tasks with a metronome set at 2.0 Hz.¹⁹ Interventions may need to be sex specific given that men with age-related vestibular loss increase their gait speed, whereas affected women decrease their speed.¹³

This study has limitations. It was a cross-sectional study, and although the effect sizes for specific gait metrics were very large, the analyses presented here were completed post hoc, and the sample size, particularly for each vestibular loss subgroup, was small. The minimal detectable change values from these preliminary results will be useful for planning future work. The group of adults with vestibular loss was older than the group of healthy adults, and this very large difference in age contributed to the very large effect sizes for GV and TD. Thus, future studies should enroll age-matched participants. The instrumented gait analysis was conducted with 3 IMUs; a more complete characterization of vestibulopathic gait is possible if additional sensors are used. Associations between persistent gait deficits and risk of falls and between IMU-based gait analysis and observational gait tests should be explored further.

These results suggest that adults with chronic UVL and BVL have enduring clinically and functionally meaningful deficits in gait stability and turning dynamics. Both SL and PTV may have good criterion validity, and turn angle, TD, and mean velocity also distinguish adults with BVL from those with UVL. Practitioners should target gait stability and turning dynamics when designing customized VR for those with chronic vestibular loss who have persistent deficits.

Accepted for Publication: May 8, 2021.

Published Online: July 1, 2021. doi:10.1001/jamaoto.2021.1276

Corresponding Author: Colin R. Grove, PT, DPT, PhD, Department of Surgery, University of Wisconsin–Madison, 6630 University Ave, Middleton, WI 53562 (crgrove@wisc.edu).

Author Contributions: Dr Grove had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: All authors.

Acquisition, analysis, or interpretation of data: Grove, Whitney, Heiderscheit.

Drafting of the manuscript: Grove.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Grove.

Obtained funding: Grove, Heiderscheit.

Administrative, technical, or material support: Grove, Heiderscheit.

Supervision: Whitney, Pyle, Heiderscheit.

Conflict of Interest Disclosures: Dr Grove reported receiving grants from the National Center for Advancing Translational Sciences during the conduct of the study and consulting fees from Wicab Inc outside the submitted work. Dr Whitney reported receiving personal fees from Intelligent Automations and MedBridge and grants from the US Department of Defense outside the submitted work. Dr Heiderscheit reported receiving personal fees from Altec Inc outside the submitted work. No other disclosures were reported.

Funding/Support: The project described was supported by the Clinical and Translational Science Awards program through grants UL1TR000427 and TL1TR002375 from the National Institutes of Health (NIH). Additional funding was provided by the University of Wisconsin through a research assistantship for Dr Grove through the Department of Surgery and a research grant from the Department of Orthopedics and Rehabilitation.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Additional Contributions: Scott J. Hetzel, MS, Department of Biostatistics, University of Wisconsin–Madison, advised Dr Grove regarding the statistical analysis. Kristen E. Caldera, DO, Department of Orthopedics and Rehabilitation, University of Wisconsin–Madison, assisted the study team in obtaining funding. Neither were compensated for their work. Physical therapists from the University of Wisconsin Hospital and Clinics assisted in recruiting participants but were not compensated for these referrals.