Participants selection

A total of 51 consecutive unilateral vestibular hypofunction participants were enrolled from February 2019 to March 2021 by the Uniter Onlus, a regional institutional interdisciplinary disorder clinic, after they had been enrolled in the local longitudinal cohort study. Only patients presenting with a concurrent reduction in vestibulo-ocular reflex gain were included in this study (see Video Head Impulse Test paragraph below).

By using a frequency-matching procedure, a population of 60 gender- and age-matched healthy patients serving as the control group were extracted from the institutional database by a biostatistician who was not involved in the study.

According to previous experiences, diagnosis of chronic unilateral vestibular hypofunction was achieved,^{Reference Hall, Herdman, Whitney, Anson, Carender and Hoppes14} using bithermal caloric irrigation, with a reduced response on one side of at least 25 per cent when calculated by means of Jongkees’ formula. Disease onset was to be at least three months earlier.^{Reference Alessandrini, Micarelli, Chiaravalloti, Candidi, Bruno and Di Pietro17} No patient had ever undergone vestibular rehabilitation treatment before.

A clinical otoneurological examination, including binocular electro-oculography analysis with positional manoeuvres, Head Shaking Test, clinical Head Impulse Test, gait observation, limb co-ordination tests, Romberg stance test, pure tone audiometry and impedance was performed on all study participants.^{Reference Micarelli, Viziano, Bruno, Micarelli and Alessandrini18} Hearing loss was accounted for in the matching procedure between unilateral vestibular hypofunction participants and the control group, according to the American Academy of Otolaryngology guidelines.¹⁹

Demographic data, such as age and gender, and anthropometric measurements for all study participants were extracted from their charts, as well as (for patients with unilateral vestibular hypofunction) disease cause, side and duration from onset (disease duration, in months).

Sleep evaluation

According to the American Academy of Sleep Medicine criteria,^{Reference Berry, Budhiraja, Gottlieb, Gozal, Iber and Kapur20} and in order to exclude sleep disorders,^{Reference Micarelli, Liguori, Viziano, Izzi, Placidi and Alessandrini11} all patients underwent home-based polygraphic cardiorespiratory monitoring in order to evaluate the Apnea–Hypopnea Index and parameters related to arterial oxygen saturation: mean arterial oxygen saturation, lowest arterial oxygen saturation, mean arterial oxygen desaturation and Oxygen Desaturation Index (number of oxygen desaturation episodes of equal to or more than 3 per cent per hour).^{Reference Berry, Budhiraja, Gottlieb, Gozal, Iber and Kapur20,Reference Liguori, Romigi, Izzi, Mercuri, Cordella and Tarquini21}

Exclusion criteria

Participants were excluded from this study if the clinical history on their charts reported cardiovascular, rheumatological, orthopaedic or neurological conditions, or previous falls. All participants were screened for blood test abnormalities, such as evidence of kidney or liver dysfunction. Pregnant or breastfeeding patients were excluded. Chronic neurological impairment was excluded by means of the Mini Mental State Examination and magnetic resonance imaging. Other abnormalities possibly causing otoneurological dysfunction, such as neuro-psychiatric disorders, lung disease, vitamin deficiencies, diabetes and hypothyroidism were screened for, and patients with positive history or examination were not included in the study. Participants taking drugs impacting on the cochleo-vestibular system or with a history of drug or tobacco or alcohol addiction were excluded. Individuals were also excluded from the study if they were unable to understand examination procedures or participate in such procedures because of other physical conditions.^{Reference Micarelli, Viziano, Carbini, Misici, Guzzo and Micarelli16}

Ethics committee and public registration

The study was approved by the Regional Lazio institutional review board (registration number: 1668/2017), and it adhered to the principles of the Declaration of Helsinki. The study was registered in the public registry ClinicalTrials.gov (number: NCT05174104), and all participants provided written informed consent after receiving a detailed explanation of the study. Participants with unilateral vestibular hypofunction, before and after vestibular rehabilitation, and the control group underwent the tests in the following sections.

Video Head Impulse Test

For Video Head Impulse Test measurements, a protocol used in previous studies was implemented.^{Reference Micarelli, Viziano, Bruno, Micarelli and Alessandrini18,Reference Blödow, Pannasch and Walther22,Reference Alessandrini, Micarelli, Viziano, Pavone, Costantini and Casali23} Results were classified as abnormal if gain was off the scale of calculated normative data and if refixation saccades (showed by visual inspection, according to Blödow et al.^{Reference Blödow, Pannasch and Walther22}) were present. With the manufacturer's software (OtoAccess™), median values from both sides, recorded at 60 ms were extracted on Excel^® spreadsheet software files for raw data analysis. Diagnosis of unilateral vestibular hypofunction, as per previous procedures,^{Reference Micarelli, Viziano, Bruno, Micarelli and Alessandrini18,Reference Blödow, Pannasch and Walther22} was confirmed with vestibulo-ocular reflex gain values below 0.75 and 0.76 for left and right sides, respectively. This was calculated as the lower value of the laboratory gain reference range regarding a population of healthy volunteers (calculated as mean_normal ± 2 (standard deviations (SDs)), which is equal to 0.89 ± 2 (0.07) and to 0.88 ± 2 (0.06) for left and right sides, respectively.

Static posturography testing

In order to monitor the centre of pressure during posturography analysis, each patient was instructed to keep an upright position on a standardised platform (ED800, Medi-Care Solutions, Euroclinic, Bologna, Italy).^{Reference Alessandrini, Micarelli, Viziano, Pavone, Costantini and Casali23} The recording period was 60 seconds for each test (eyes closed or open while standing on the stiff platform), and the sampling frequency in the time domain was 25 Hz.^{Reference Alessandrini, Micarelli, Viziano, Pavone, Costantini and Casali23–Reference Micarelli, Viziano, Della-Morte, Augimeri and Alessandrini25} The centre of pressure was monitored while performing the test. The posturographical parameters considered in our study were the path length (in millimetres) and the 95 per cent confidence ellipse area (area in millimetres squared), both calculated according to Prieto et al.^{Reference Prieto, Myklebust, Hoffmann, Lovett and Myklebust26}

Actigraphy-based sleep analysis

Actigraphy analysis was performed for a period of one week by means of the wGT3X-BT device by ActiGraph (Pensacola, USA). Such a device is a triaxial accelerometer measuring wrist acceleration in three orthogonal axes at a sampling frequency of 80 Hz. This device is waterproof, with a battery life of approximately three weeks. The device did not provide any feedback to the participants about their activity or sleep.^{Reference Asgari Mehrabadi, Azimi, Sarhaddi, Axelin, Niela-Vilén and Myllyntausta27} Data collected by the device were utilised to estimate the following sleep parameters: sleep onset latency, total sleep time, wake time after sleep onset, and sleep efficiency for each participant. These parameters were calculated for the whole week period.^{Reference Micarelli, Viziano, Pistillo, Granito, Micarelli and Alessandrini3,Reference Martin, Moussay, Bulla, Bulla, Toupet and Etard28}

Raw data were converted into 60-second epochs by means of manufacturer-provided software (ActiLife software version 6.13, ActiGraph).^{Reference Asgari Mehrabadi, Azimi, Sarhaddi, Axelin, Niela-Vilén and Myllyntausta27} Non-wear periods were excluded based on criteria followed in previous experiences.^{Reference Troiano, Berrigan, Dodd, Mâsse, Tilert and McDowell29} Sleep data were systematically checked for accuracy by the same researcher, with the subsequent exclusion of obvious outliers.^{Reference Asgari Mehrabadi, Azimi, Sarhaddi, Axelin, Niela-Vilén and Myllyntausta27} According to recommendations,^{Reference Asgari Mehrabadi, Azimi, Sarhaddi, Axelin, Niela-Vilén and Myllyntausta27,Reference Ancoli-Israel, Cole, Alessi, Chambers, Moorcroft and Pollak30} both groups’ patients were also asked to report their sleep times, such as bedtime, wake-up time and naps, during the seven-day study period via a self-report form, structured like a daily log. Patients were also instructed to report other relevant events happening during the study recording period, such as device removal from the wrist or clinical symptoms. The self-report data were used as an aid and an integration in the interpretation of actigraphy data, thus mitigating possible errors.

Self-report and performance measures

The Italian Dizziness Handicap Inventory version was used for this study. It comprises 25 items designed to assess functional (9 questions), emotional (9 questions) and physical (7 questions) limitations on a three-point scale and their sum (total Dizziness Handicap Inventory).^{Reference Nola, Mostardini, Salvi, Ercolani and Ralli31}

The Activities-Specific Balance Confidence scale was used in order to assess patient perceived level of balance confidence during 16 daily living activities, with scores ranging from 0 to 100 per cent.^{Reference Herdman, Hall, Maloney, Knight, Ebert and Lowe32}

The Dynamic Gait Index was used to evaluate the ability to perform various gait activities, such as walking with head turns and avoiding obstacles.^{Reference Herdman, Hall, Maloney, Knight, Ebert and Lowe32} The scale has eight items, and each item is scored from zero to three.^{Reference Herdman, Hall, Maloney, Knight, Ebert and Lowe32}

Pittsburgh Sleep Quality Index is a self-rated questionnaire about sleep quality during the last month, consisting of 19 items with a 4-point Likert-type response format.^{Reference Curcio, Tempesta, Scarlata, Marzano, Moroni and Rossini33} Pittsburgh Sleep Quality Index contains a zero to three interval scale with seven components (subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep difficulty, use of sleeping medication and difficulty of diurnal awakening). A cut-off score of five has shown good sensitivity and specificity for people with sleep disorders.^{Reference Curcio, Tempesta, Scarlata, Marzano, Moroni and Rossini33}

The Italian version of the reduced Morningness–Eveningness Questionnaire was used to evaluate the patient circadian rhythm.^{Reference Natale, Esposito, Martoni and Fabbri34} This scale contains five items, related to preferred bedtime, get-up time, tiredness in the morning, peak performance and a global self-assessment item. Based on their scores, individuals were categorised as being evening type (4–10), intermediate type (11–18) and morning type (19–25).^{Reference Natale, Esposito, Martoni and Fabbri34}

Vestibular rehabilitation protocol

In light of diagnostic test results, history and physical examination, a personalised exercise programme was developed by an otoneurologist and administered to patients by a skilled physiotherapist, who was blinded to the study. The treatment programme consisted of training and exercise components. The rationale and clinical basis, as well as contraindications for performing any exercise were explained during the training component. Patients affected by unilateral vestibular hypofunction were actively involved in adapting the exercise programme to suit their symptoms, capabilities and lifestyle. The execution of such exercises was personalised as much as possible by the therapist, based on individual symptoms and disability. The exercises were designed to be challenging during the training period, and different aspects of balance training were emphasised for different participants to provide further individualisation. Exercises can be summarised as follows.^{Reference Herdman, Hall, Maloney, Knight, Ebert and Lowe32,Reference Giray, Kirazli, Karapolat, Celebisoy, Bilgen and Kirazli35}

Data handling and statistical analysis

Following previous experiences strongly associating unilateral vestibular hypofunction with balance control alterations and to changes in self-report measures or performance measures,^{Reference Micarelli, Viziano, Augimeri, Micarelli and Alessandrini15,Reference Micarelli, Viziano, Bruno, Micarelli, Augimeri and Alessandrini36} the sample size was calculated to detect inter-group differences in the results for the posturography variables (surface and length in both closed and open eye conditions) and Dynamic Gait Index, Dizziness Handicap Inventory and Activities-Specific Balance Confidence scale.

The sample size for the test hypothesis was calculated accordingly with the context (independent samples and continuous variables), using a statistical power of 80 per cent (1-β) for an error probability of 0.05. We used the t-test for independent samples and an effect size of 0.80. We thus determined to include at least 44 participants per group.

The chi-squared test was carried out to define associations between categorical factors (including Pittsburgh Sleep Quality Index and reduced Morningness–Eveningness Questionnaire categorisations) and groups. Given their quantitative nature, descriptive data were calculated as mean and SD for otoneurological values, self-report and performance measure scores, and actigraphy measurements. In order to assess that data for independent samples were of Gaussian distribution, D'Agostino's K-squared normality and Levene's homoscedasticity test were applied (where the null hypothesis is that the data are normally and homogeneously distributed).

A between-group analysis of variance was performed for each otoneurological, self-report or performance measure score and actigraphy variable. In order to obtain a unique entry of the vestibulo-ocular reflex gain, the average between right and left side was calculated in control group participants in agreement with previous quantitative studies.^{Reference Micarelli, Liguori, Viziano, Izzi, Placidi and Alessandrini11,Reference Moon, Chang and Kim37} Gender was treated as a categorical predictor, and age and unilateral vestibular hypofunction disease duration were treated, where possible, as continuous predictors. The significant cut-off level (α) was set at a p-value of 0.05. Bonferroni correction for multiple comparisons was used for the post hoc test of the significant main effects, and the corrected level of significance was set at 0.017. Then, according to previous protocols,^{Reference Micarelli, Viziano, Della-Morte, Augimeri and Alessandrini25} and given the exploratory nature of the study, a two-tailed Spearman's rank correlation was performed in unilateral vestibular hypofunction participants between pre- and post-treatment differences (change) in those otoneurological, self-report and performance measures and actigraphy measurement scores that were significantly different when comparing the pre- and post-scores. Given the large sample size of this group and the two-tailed nature of the test, a significant cut-off level (α) was set at a p-value of 0.05 (Statistica^® 7 package for Windows).