Participants
Participants were recruited at a tertiary referral center (Maastricht UMC +) between June 2016 and December 2018. Inclusion criteria for BVP-patients were in accordance with the Diagnostic criteria Consensus document of the Classification Committee of the Bárány Society [22]: a horizontal angular VOR gain on both sides < 0.6 (angular head velocity 150–300°/s) and/or summated slow phase velocity of nystagmus of less than 6°/s on each side during bithermal caloric tests (30 and 44 °C, 300 ml in 30 s) and/or a horizontal angular VOR gain < 0.1 upon sinusoidal stimulation on a rotatory chair (0.1 Hz, Vmax = 50°/s) and/or a phase lead > 68° (time constant of < 5 s). In addition, patients needed to be older than 18 years. Patients unable to stop medication against anxiety or depression 1 week before testing, were excluded from the study, as well as those suffering from peripheral neuropathy of the lower extremities.
Healthy subjects were recruited via posters in the hospital and among families and friends of the researchers. A questionnaire was used to rule out, as much as possible, any deficits or diseases that could influence the vestibular system. It comprised the following topics: previous medical history (including otorhinolaryngological, neurological, ophthalmological); use of any medication; known balance problems, recent neck trauma or dizziness in the past 6 months.
All participants were excluded from the study if they were unable to walk on the treadmill at 2 km/h or had a vision of − 4.0 Diopter or lower (without correction), in which they could not read the first line of the optotypes on the computer screen. In some cases, BVP-patients were allowed to hold handrails of the treadmill to prevent falling (10 BVP-patients). The use of alcohol or other stimulants was forbidden in the 24 h before examination.
Evaluating age effect on the vestibular function in BVP-patients
The potential effect of age on vestibular function in BVP-patients (which could affect drop-out rate and DVA) was assessed using outcomes from two tests: the video Head Impulse Test (vHIT, ICS Impulse, GN Otometrics; Taastrup, Denmark) and the caloric test using bithermal (30° and 40° C) irrigations of water. vHIT gains were calculated for the leftward and rightward directions in the lateral plane, and for the upward and downward directions in right anterior—left posterior and left anterior—right posterior planes. The sum of bithermal maximum peak slow phase velocities (SPV) of the nystagmus was used as outcome measure of the caloric test, calculated separately for each side.
Testing DVA on a treadmill
Sloan optotypes (C, D, H, K, N, O, R, S, V and Z) projected on a computer screen were used to test visual acuity. The computer screen was placed at eye level and at 2.8 m from the subject. A Sloan letter was presented on a computer screen (LG 24bk55 24″), using a custom program written in Matlab R2010a (The Mathworks, Natick, MA, USA). The program randomly showed five letters in a single-letter sequence at one logarithm of the Minimum Angle of Resolution (logMAR). If a minimum of two out of five letters were correctly recognized, the letter size decreased by 0.1 logMAR and five new letters were shown, one after another. If less than two out of five letters were recognized, the procedure was stopped.
The visual acuity of all BVP-patients and healthy subjects was tested in two conditions: static and dynamic. Static visual acuity was measured when the subject was standing still on the treadmill (1210 model, SportsArt, Inc., Tainan, Taiwan). Visual acuity in dynamic conditions was measured while walking on the treadmill at different speeds (2, 4, 6 km/h, non-randomized). The study procedure ended either when all walking speeds were completed, or when subjects could not walk at a higher speed. If subjects were not able to complete the test at a specific walking speed, they were considered as a “drop-out” for that walking speed. To ensure the subject’s safety, a safety string was clipped to the subject’s waist that was connected to the emergency brake of the treadmill.
Data analysis and statistics
Visual acuity in static and dynamic conditions was calculated as: LogMAR = 0.1 + LogMARx − 0.02y [23], where “x” was defined as the last optotype line in which two or more letters were read correctly and “y” was defined as the number of correctly read letters at that line.
DVAL was defined as the difference between visual acuity in the static condition and the dynamic conditions. Note that a negative DVAL indicates poorer vision in the tested dynamic condition compared to the static condition. Descriptive statistics were made for age and DVAL. The independent sample t test was used to compare mean age between groups.
To evaluate whether holding the treadmill handrails affected DVAL, an independent t test was used to assess whether a difference in mean DVAL existed between patients who held the treadmill handrails and patients who did not hold the treadmill handrails. Obtained p values were Bonferroni corrected.
The potential effect of age on vestibular function in BVP-patients was analyzed using linear regression analyses. Each model contained age as an independent variable and the corresponding outcome (gain or SPV) as a dependent variable.
Since drop-out at the speed of 4 and 6 km/h was perfectly correlated (e.g., drop-out at 4 km/h excluded successful completion of the test at 6 km/h), multilevel logistic regression of drop-out (yes or no) on speed, age and group (BVP-patients or healthy subjects) failed (multicollinearity). Therefore, a new dependent variable reflecting the missing pattern was formed using the following criteria: pattern 1—no drop-out at all speeds, pattern 2—drop-out at 6 km/h, and pattern 3—drop-out at 4 and 6 km/h. Then multinomial logistic regression was performed to determine the dependency of drop-out on group and age.
To analyze the effect of age and speed on DVAL, while accounting for the dependence among measurements of the same participant, a linear-mixed effects model was applied. Initially, age, group, speed, and all their two-way interactions were included as fixed factors. Then, the non-significant interactions were removed by backward selection. Finally, age, group, speed and group by speed interaction were left in the model. Pairwise comparisons were made per group to compare DVAL at 2, 4, and 6 km/h. Pairwise comparisons were also made per speed to compare DVAL in BVP-patients and healthy subjects. The significance level, alpha, was set to 0.05. In case of multiple comparisons, p values were Bonferroni corrected. Data were analyzed in R (v.3.5.2) and SPSS (v.25).
Ethical considerations
This study was in accordance with the Declaration of Helsinki (amended version 2013) Approval was obtained by the ethical committees of Maastricht University Medical Centre (NL52768.068.15/METC 151027). All participants provided written informed consent prior to the study.