Online vestibular rehabilitation for chronic vestibular syndrome: 36-month follow-up of a randomised controlled trial in general practice

Background Vestibular rehabilitation (VR) is the preferred treatment for chronic vestibular symptoms such as dizziness and vertigo. An internet-based programme was developed to increase uptake of VR. The authors have previously reported that internet-based VR resulted in a clinically relevant decrease of vestibular symptoms for up to 6 months, compared with usual care. Aim To evaluate long-term outcomes of internet-based VR in patients with chronic vestibular syndrome. Design and setting A randomised controlled trial was conducted in Dutch general practice involving 322 participants aged ≥50 years with chronic vestibular syndrome. Participants were randomised to stand-alone VR, blended VR (with physiotherapy support), and usual care. Usual care participants were allowed to cross over to stand-alone VR 6 months after randomisation. Method Participants were approached 36 months after randomisation. The primary outcome was the presence of vestibular symptoms as measured by the vertigo symptom scale—short form (VSS–SF). Secondary outcomes were dizziness-related impairment, anxiety, depressive symptoms, and healthcare utilisation. Results At 36-month follow-up, 65% of participants filled in the VSS–SF. In the usual care group, 38% of participants had crossed over to VR at 6 months. There were no significant differences in vestibular symptoms between VR groups and usual care (mean difference = −0.8 points, 95% confidence interval [CI] = −2.8 to 1.2, for stand-alone VR; −0.3, 95% CI = −2.2 to 1.7, for blended VR). In VR groups, clinically relevant improvement compared with baseline was maintained over time. Conclusion Internet-based VR provides a maintained improvement of vestibular symptoms for up to 36 months in patients with chronic vestibular syndrome.


INTRODUCTION
GPs often encounter patients with vestibular symptoms, such as dizziness and vertigo. 1 In the general population the lifetime prevalence of vestibular symptoms varies between 17% and 30%. 2,3Older patients generally experience more frequent and severe symptoms than younger patients and have an increased chance of developing chronic symptoms (known as chronic vestibular syndrome in the current nomenclature). 2,4,57][8] VR consists of specific exercises designed to maximise central nervous system compensation after vestibular disorders occur.In addition, a form of exposure-based behaviour therapy takes place by provoking vestibular symptoms in a controlled context. 9,10To increase uptake of VR, an internet-based intervention was developed.In 2017, this online treatment was first shown to be effective in a randomised controlled trial in the UK by researchers at the University of Southampton. 10,11The authors of the present article worked together with researchers at the University of Southampton to create a Dutch version of internet-based VR and subsequently conducted a large, randomised controlled trial in the Netherlands investigating the effectiveness of internet-based VR both without (stand-alone VR) and with support (blended VR). 12 The authors previously reported that, compared with usual care, internet-based VR decreased vestibular symptoms for up to 6 months in patients with chronic vestibular syndrome in Dutch primary care. 13In this article, the authors report the 36-month outcomes.

METHOD Study design
A pragmatic, three-armed, individually randomised controlled trial was conducted involving 59 Dutch general practices.The Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines for non-drug treatment interventions were followed and the trial was registered in the Netherlands Trial Register (reference: NTR5712).A detailed research protocol and the results of the short-term effectiveness analysis at 6 months were published previously. 12,13Participants were recruited between June 2017 and July 2018.For this pre-specified long-term 36-month follow-up, participants were approached again 3 years after randomisation between August 2020 and July 2021.

Participants
Electronic medical records were used to identify patients aged ≥50 years who visited e710 British Journal of General Practice, September 2023 their GP with a vestibular symptom in the past 2 years.The GP screened potentially eligible participants to exclude those with medical contraindications for making the required head movements, serious comorbid conditions that precluded participation in an exercise programme, an identifiable non-vestibular cause of their symptoms, or current enrolment in a related study.Trial information and a form to express interest was provided to potential participants.Physicians in the research team checked the eligibility criteria of interested patients by telephone.The inclusion criteria were: • good command of the Dutch language; • access to the internet and an email account; • persisting vestibular symptoms at time of inclusion, present for ≥1 month; and • vestibular symptoms exacerbated or triggered by head movements.
These inclusion criteria were used to identify participants with a chronic vestibular syndrome, as defined by the Bárány Society in the International Classification of Vestibular Disorders, 5,14 who were deemed suitable to receive online VR (Supplementary Table S1).

Randomisation and blinding
After the informed consent procedure, participants were sent an email with a link to the trial website.On completion of the baseline questionnaire, the software program allocated them to internet-based VR (stand-alone VR), internet-based VR with support (blended VR), or usual care.The randomisation process was fully automated and concealed from the research team.Due to the nature of the trial interventions, it was not possible to blind participants, physiotherapists, and research assistants.Vertigo Training was a Dutch translation of the safe and effective British internetbased VR intervention Balance Retraining (freely available from https://balance.lifeguidehealth.org). 11Vertigo Training lasted 6 weeks and consisted of weekly online sessions and daily VR exercises, explained in more detail in the previously published short-term effectiveness analysis 13 and Appendix (Supplementary Table S2).Stand-alone VR participants received the standard level of care from their own GP without any restrictions.

Interventions
Blended internet-based VR with physiotherapy support.Blended VR participants were granted access to the same online Vertigo Training intervention as the stand-alone VR participants.In addition, they were visited twice at home by a trained physiotherapist during the 6-week intervention period.These 45-minute physiotherapy sessions took place in the first and third week.The physiotherapist provided information about the background of vestibular symptoms and VR, and talked about doubts and concerns the participant might have.They also taught participants how to use the online intervention, took participants through a set of VR exercises, advised on coping with obstacles to adherence, and encouraged participants to continue the exercises.Blended VR participants also received the standard level of care from their own GP with no restrictions.
Usual care.Usual care participants received standard care from their own doctor without restrictions.The authors provided participating doctors with written instructions, asking them to diagnose and treat all causes of vestibular symptoms according to the guidelines of the Dutch College of General Practitioners. 15After completing the short-term follow-up measures at 6 months, all usual care participants were offered access to stand-alone VR.Usual care participants who chose to cross over at this time were registered in a separate database.

How this fits in
Vestibular rehabilitation (VR) is an exercise-based therapy that can be used to decrease vestibular symptoms, such as dizziness and vertigo, by gradually stimulating the vestibular system.Although VR is the preferred treatment for chronic vestibular syndromes, the intervention is still largely underused in clinical practice.In a randomised controlled trial in general practice, internet-based VR was previously shown to result in a clinically relevant decrease of vestibular symptoms over 6 months.In this 36-month follow-up study, the authors found no significant differences in vestibular symptoms between internet-based VR groups and usual care, but clinically relevant effects after VR were maintained over time.

Primary and secondary outcomes
Measurements were collected at baseline, 3-, 6-, and 36-months' follow-up.At baseline, participants provided information on their age, sex, level of education, living situation, comorbidities, vestibular diagnosis, frequency and average duration of vestibular symptoms, and the time since their vestibular disorder was diagnosed.The primary outcome was vestibular symptoms as measured by the vertigo symptom scale-short form (VSS-SF). 16,17 The questionnaire can be used to measure the frequency and severity of 15 vestibular symptoms, with more points indicating more frequent/severe symptoms (range 0-60).9][20] The questionnaire has high internal consistency (Cronbach's α = 0.90), excellent discriminative ability (area under the curve = 0.87), and high test-retest reliability (intraclass correlation coefficient = 0.88). 17The secondary outcomes were the Dizziness Handicap Inventory questionnaire (DHI) for dizziness-related impairment; 21 and the Patient Health Questionnaire (PHQ) 22 for severity of anxiety (Generalised Anxiety Disorder Assessment [GAD-7] subscale) 23 and depressive symptoms (PHQ-9 subscale). 24For the 36-month follow-up, all primary and secondary outcome measures were assessed again, except for the single dichotomous item of subjective improvement compared with baseline.In addition, enquiries were made about the participants' healthcare use and long-term adherence to the exercises.Participants were asked if they had visited any healthcare professional for their vestibular symptoms since the last follow-up measure.The stand-alone VR and blended VR participants were also asked how long they continued using the VR exercises after receiving access.Usual care participants who chose to cross over to stand-alone VR at 6 months were registered in a separate database.

Statistical analysis
An intention-to-treat analysis was performed to compare stand-alone VR and blended VR versus usual care.A linear mixed-model analysis was used to account for repeated measures within individuals.In a longitudinal dataset, this technique accounts for missing data without performing multiple imputations. 25This analysis included group assignment, time, and the interaction between group assignment and time.By creating dummy variables, time and group assignment were included as categorical variables, which enabled the calculation of outcomes for each follow-up moment while adjusting for baseline values. 26The same potential confounders were adjusted for as in the 6-month analysis. 13Because part of the usual care group started standalone VR at 6 months, an additional post-hoc analysis was conducted to assess outcomes of cross-over participants separately from other usual care participants at different time points.The same variables were used for this analysis as for the main analysis.Several post-hoc sensitivity analyses were also conducted to assess non-response bias and attrition.For non-response, differences in characteristics were descriptively analysed between participants who filled in the 36-month follow-up measurement and those who did not.A complete case analysis was also performed by repeating the primary and secondary outcome analysis without participants who did not fill in the 36-month follow-up measure.Lastly, to further assess the effects of attrition, a per-protocol analysis was completed with participants who filled in the 36-month follow-up.This included only the participants assigned to stand-alone VR who completed all six online sessions, and the participants assigned to blended VR who completed all six online sessions and both physiotherapist visits.Stata (version 14.1) was used for statistical analyses.

Secondary outcomes
There were no significant differences in impairment from dizziness between VR groups and usual care at 36 months (Table 3).
However, similar to vestibular symptoms, improvement VSS-SF = vertigo symptom scale-short form.
e714 British Journal of General Practice, September 2023 of dizziness- related impairment was maintained at 36 months in the stand-alone VR (-9.9 points, 95% CI = -13.3 to -6.4) and blended VR groups (-10.4,95% CI = -13.3 to -7.5).In the post-hoc analysis, usual care participants who crossed over to VR at 6 months showed an improvement of dizziness-related impairment thereafter (6 months: -6.6 points, 95% CI = -12.8 to -0.4; 36 months: -11.0, 95% CI = -17.3 to -4.7), whereas participants who did not cross over spontaneously improved at 6 months and maintained that improvement at 36 months (6 months: -10.1 points, 95% CI = -14.5 to -5.7; 36 months: -10.8, 95% CI = -15.2 to -6.4).There were no significant differences in anxiety and depressive symptoms between VR groups and usual care after 36 months.Significant improvement of anxiety symptoms was maintained over time in both VR groups, while improvement of depressive symptoms was only maintained in the blended VR group.In the usual care group significant improvement of depressive symptoms was seen at 36 months, but not at previous time points.Additionally, participants were asked about their healthcare use in the period between the 6-month and 36-month follow-up (Table 4).No relevant differences were detected between VR groups and usual care in visits to the GP, GP out-of-hours service, medical specialist, emergency department, and hospital admissions for vestibular symptoms.A few patients were admitted to the hospital (stand-alone VR, n = 4 and usual care, n = 1).Reasons for admission were 'transient ischaemic attack and epileptic attack' (n = 1), 'fall' (n = 1), 'atrial fibrillation' (n = 1), and 'not reported' (n = 2).No patient mentioned adverse events related to VR exercises.
Lastly, participants were asked to elaborate on their adherence to VR after the trial (Table 4).Even though the treatment lasted for 6 weeks, 18% of participants continued to use VR regularly up to 3 months after start of the trial and 16% of the participants used VR regularly for >12 months.At 36 months, 63% of the participants who completed the questionnaire still continued to perform VR exercises.

Post-hoc sensitivity analyses for non-response bias and attrition
To assess non-response bias and attrition the authors conducted several sensitivity analyses.The baseline characteristics for participants who filled in the 36-month follow-up and those who did not are described separately in Supplementary Table S3.Participants who completed 36-month follow-up had a relatively higher level of education, were less likely to live alone, and more often followed the per-protocol treatment.The results of a subsequent complete case sensitivity analysis, where only the patients who filled in the 36-month were included, showed similar results to the primary analysis (Supplementary Table S4).Lastly, a perprotocol analysis with participants who filled in the 36-month follow-up showed that VR participants who followed all sessions had fewer vestibular symptoms at long-term follow-up, indicating a positive effect of better engagement (Supplementary Table S5).

Summary
At 36-month follow-up, vestibular symptoms did not significantly differ between participants who received internet-based VR and usual care.The same applies with regard to impairment experienced due to dizziness, anxiety, and depressive symptoms.Nevertheless, clinically relevant improvements in internet- based VR groups were maintained over time.Recovery of usual care participants who crossed over at 6 months further indicates the value of online VR when vestibular symptoms do not improve spontaneously.There were no clear differences in healthcare utilisation between groups at long-term follow-up, and serious adverse events due to VR were unlikely.Many patients reported continuing VR exercises long after the end of the 6-week intervention period.

Strengths and limitations
This study has several strengths.This was a large, well-designed trial and participants showed good adherence to the intervention. 13Furthermore, the primary outcome, the VSS-SF, is a validated patientreported outcome measure that is used in most major VR trials. 17,27Another strength is the high response rate in this long-term follow-up in a sample of older patients that can be prone to attrition.There are also some limitations.The response rate at 6-month follow-up was higher than at 36-month follow-up.The number of participants at 36-month followup was lower than the 80 participants per group in the original power calculation for 6-month follow-up. 13Post-hoc power analyses are generally not considered useful. 28Because this is the first long-term follow-up of VR, to the authors' knowledge, it is also impossible to estimate the expected treatment effect.However, the lower long-term follow-up rate may have made it harder to detect significant differences.The results presented in this long-term follow-up may not be representative for all, as the post-hoc sensitivity analyses showed that participants who were lost to follow-up were relatively lower educated and more likely to live alone.It is therefore important to pay special attention to these groups.The per-protocol analysis showed that following all online VR sessions may positively influence results, so finding ways to improve engagement for all will be vital to successful implementation.Another limitation for the comparison between groups was the cross over of participants in the usual care group.The decision to allow participants to cross over was made before the trial to increase the benefits of inclusion for participants, but it complicated the comparison between VR groups and usual care in this long-term follow-up.Lastly, this study did not collect data on specific treatments for dizziness received by participants during the follow-up period.Nevertheless, given the relative absence of effective treatments available in chronic vestibular syndrome, it is unlikely that this strongly influenced results.
Stand-alone internet-based VR.Standalone VR participants received access to the Vertigo Training internet-based intervention.

Figure 2 .
Figure 2. Primary outcome 36-month follow-up results after VR.VSS-SF range is 0-60, higher values indicate more severe/frequent vestibular symptoms, clinically relevant difference ≥3 points (indicated by dotted black line).Primary analysis: all participants analysed according to allocation (total, N = 322; stand-alone VR, n = 98; blended VR, n = 104; and usual care, n = 120).Post-hoc analysis: usual care at 3-year follow-up (n = 76) andcomparison of participants who did not cross over to stand-alone VR after 6 months (n = 47) versus those who did (n = 29).All analyses were adjusted for baseline VSS-SF values, repeated measurements within participants' age, gender, level of education, living situation, number of chronic diseases, time since vestibular diagnosis, and the presence of a panic disorder, generalised anxiety disorder, or major depressive disorder at baseline.

Table 2 . Comparison primary outcome between treatment groups at 36 months Comparison, adjusted mean difference Change from baseline, mean (95% CI) a
a Unless otherwise stated.bAdjustedforbaselinevalues, repeated measurements within participants' age, gender, level of education, living situation, number of chronic diseases, time since vestibular diagnosis, and the presence of a panic disorder, generalised anxiety disorder, or major depressive disorder at baseline.cPrimaryanalysis: all participants analysed according to allocation (total, N = 322; stand-alone VR, n = 98; blended VR, n = 104; and usual care, n = 120).d Post-hoc analysis: usual care at 3-year follow-up (n = 76) and comparison of participants who did not cross over to stand-alone VR after 6 months (n = 47) versus those who did (n = 29).SD = standard deviation.VR = vestibular rehabilitation.

Table 3 . Comparison of secondary outcomes between treatment groups at 36 months Comparison, adjusted mean difference Change from baseline, mean (95% CI) a (95% CI) a,b
a Unless otherwise stated.b Adjusted for baseline values, repeated measurements within participants age, gender, level of education, living situation, number of chronic diseases, time since vestibular diagnosis, and the presence of a panic disorder, generalised anxiety disorder, or major depressive disorder at baseline.c Primary analysis: all participants analysed according to allocation (total, N = 322; stand-alone VR, n = 98; blended VR, n = 104; and usual care, n = 120).d Post-hoc analysis: usual care at 3-year follow-up (n = 76) and comparison of participants who did not cross over to stand-alone VR after 6 months (n = 47) versus those who did (n = 29).DHI = dizziness handicap inventory.GAD-7 = generalised anxiety disorder assessment-7.PHQ-9 = patient health questionnaire-9.SD = standard deviation.VR = vestibular rehabilitation.e716 British Journal of General Practice, September 2023