Effects of adding transcutaneous vagus nerve stimulation to vestibular rehabilitation on dizziness and postural control in vestibular migraine: a randomized clinical trial

Abstract

Vestibular migraine (VM) causes episodic dizziness and imbalance, yet treatment options remain limited. Although management often follows migraine protocols, vestibular and headache symptoms may not coincide. While vestibular rehabilitation (VR) is a noninvasive approach to improve balance, transcutaneous auricular vagus nerve stimulation (taVNS) has shown promise in migraine relief and may influence vestibular function. This parallel-group randomized clinical trial investigated the combined effects of VR and taVNS on postural control and migraine-related symptoms in VM patients. Twenty-eight participants were randomly assigned to four groups: VR, taVNS, VR plus sham-taVNS, and VR plus taVNS. The one-month intervention included daily VR exercises and twelve 30-minute taVNS sessions targeting the cymba conchae. Outcomes comprised postural control parameters under four conditions as primary outcomes, and video head impulse test gain, Dizziness Handicap Inventory, Vertigo Symptom Scale–short form, vertigo/dizziness and headache frequency, and Headache Impact Test (HIT-6) scores as secondary outcomes. All groups showed significant improvement (p < 0.05), but VR plus taVNS produced the most consistent gains across postural and symptom measures. Groups receiving taVNS had greater reductions in headache frequency and HIT-6 scores (p < 0.05). Postural improvement moderately correlated with decreased vertigo/dizziness frequency. Combining taVNS with VR may enhance balance and symptom relief in VM.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-39697-9.

Introduction

Vestibular migraine (VM) is the most prevalent neurological cause of vertigo¹. It is considered the leading cause of spontaneous episodic vertigo, with an estimated prevalence of approximately 1% in the general population². In addition, VM affects 11% of patients in specialized vertigo clinics² and 13% of those in headache clinics³. Like migraine, VM is more common in women. Symptom onset has been reported as early as age 8 and as late as over 50, with a peak between the mid-30s and mid-40s⁴.

The term “vestibular migraine” has been recognized by the International Classification of Headache Disorders (ICHD) as a diagnostic entity encompassing both vestibular and migraine symptoms⁵. Diagnosis requires a current or past history of migraine, with or without aura, accompanied by vestibular symptoms lasting from 5 min to 72 h. At least half of the vestibular episodes must be associated with one of the following migraine features: a headache with at least two of four characteristics (unilateral location, pulsating quality, moderate to severe intensity, or worsening with routine physical activity), photophobia, phonophobia, or visual aura. In addition, other causes of vestibular symptoms must be ruled out⁵. Patients may experience dizziness, vertigo, or imbalance. In more severe cases, nausea, vomiting, motion sensitivity (visual vertigo), and postural instability may also be present⁴. Spontaneous vertigo is the most common vestibular symptom, followed by positional vertigo, vertigo induced by head movements, and vertigo triggered by visual stimuli. The duration of vestibular symptoms varies widely from a few minutes to several days⁴.

Currently, the exact pathophysiology of VM remains unclear, as the wide range of clinical and laboratory findings do not point to a single cause. Several mechanisms have been proposed, many inferred from other migraine types. Altered neural activity in the trigeminovascular system is believed to be the main mechanism responsible for headaches in migraine patients⁶.

At present, no definitive treatment exists for acute VM attacks, and management is based on general migraine guidelines. However, the association between vestibular and headache symptoms in acute VM is inconsistent: attacks may occur without headache, and when headache is present, it is often less disabling than the vestibular symptoms⁷. While analgesics can alleviate migraine headaches, they may be less effective for VM. Triptans may help in some cases, though evidence remains inconclusive⁸. Vestibular suppressants (e.g., benzodiazepines) and antiemetics (e.g., promethazine) can be used, but these often induce sedation and interfere with daily functioning⁷.

Due to the side effects of medications and the reported benefits of vestibular rehabilitation (VR) on vertigo severity in adults⁹, there is growing interest in using VR, a noninvasive intervention, for both children and adults with VM. Vestibular symptoms are biologically linked to headaches¹⁰, and interactions between vestibular and trigeminal pathways have been observed in brainstem regions such as the parabrachial nuclei, raphe, and locus coeruleus, which are involved in pain processing¹¹. VR has been shown to reduce the frequency and severity of headaches in adults with VM and is a useful clinical tool for managing migraine-associated vertigo¹². Nevertheless, there remains an urgent need for effective treatments for VM attacks.

Migraine has been associated with autonomic nervous system imbalance. Enhancing parasympathetic activity, especially via the vagus nerve, can relieve migraine symptoms¹³. Noninvasive vagus nerve stimulation (nVNS), delivered via auricular (taVNS) or cervical branches, has shown safety, efficacy, and good tolerability in treating migraine and cluster headaches^13,14. It likely exerts analgesic and anti-inflammatory effects through modulation of the trigeminovascular system via projections from the nucleus tractus solitarius (NTS)^14–16. Functional neuroimaging studies have shown that nVNS activates multiple brain regions involved in vestibular and pain processing, including the NTS, parabrachial nucleus, supramarginal gyrus, posterior insula, vestibular nuclei, cerebellum, medial temporal cortex, thalamus, medial longitudinal fasciculus, and pons¹⁷. These regions overlap with areas activated by vestibular stimulation¹⁸.

Furthermore, nVNS has shown clinical benefits in improving balance and quality of life in patients with persistent postural-perceptual dizziness (PPPD), without significant adverse effects¹⁹. The electrophysiological changes induced by nVNS in specific brain regions and in the trigeminal system may underlie its potential efficacy in treating VM²⁰. To date, only one study has examined nVNS effects in VM patients. That study assessed the effect of cervical nVNS on acute vertigo during a single session and lacked a sham control²⁰.

Although nVNS has been extensively studied in neurology, psychiatry, and rehabilitation, its application in otolaryngology remains relatively limited. Most otolaryngology-related studies have focused on otologic and related conditions, including Meniere’s disease, chronic tinnitus, and globus pharyngeus. Recent randomized and pilot clinical studies have demonstrated that taVNS can alleviate vertigo symptoms in Meniere’s disease when combined with standard medical therapy, modulate neural activity and symptom severity in chronic tinnitus, and improve symptom perception in patients with globus pharyngeus^21–23. These emerging findings suggest that taVNS may modulate auditory–vestibular and autonomic pathways relevant to vestibular disorders, providing a rationale for its investigation in vestibular migraine.

Given the positive effects of VR and the potential benefits of nVNS in VM, it is plausible that combining both therapies could produce synergistic effects. The current study aimed to evaluate the effect of adding taVNS across multiple sessions to VR on dizziness, vertigo, and postural control in patients with vestibular migraine, a combination not previously studied. We hypothesized that this approach would result in greater improvements in dizziness, vertigo, and postural control in patients with VM.

Results

The analysis was conducted using a per-protocol approach, which included the 28 participants (n = 7 per group) who completed all allocated interventions and assessments. The overall mean adherence rate for the exercises across the three VR-containing group (VR, VR+sham-taVNS, and VR+taVNS; n = 21) was 96.50 ± 3.72%. All participants included in the final analyst met the pre-defined 80% adherence threshold (at least 24 out of 30 days). The participants’ demographic characteristics, including age, sex, and the duration of vertigo/dizziness and headache, did not differ significantly between groups at baseline (Table 1). Similarly, there were no significant between-group differences in pre-intervention scores for DHI, VSS-sf, frequency of vertigo/dizziness and headache, HIT-6 scores, postural control parameters, or VOR gain (p > 0.05).

Table 1.

Demographic characteristics of the participants.

Characteristics	VR	taVNS	VR+sham-taVNS	VR+taVNS	p
age	39.71 ± 4.27	40.29 ± 8.28	41.71 ± 5.76	39.86 ± 7.44	0.939*
Sex (male/female)	2/5	1/6	1/6	2/5	0.838**
Dizziness/vertigo duration (months)	25.86 ± 17.74	24.57 ± 24.87	36.57 ± 40.97	68.00 ± 70.28	0.236*
Migraine headache duration (months)	87.00 ± 60.125	128.86 ± 79.02	168.00 ± 78.99	116.00 ± 68.82	0.240*

Values are mean ± SD, p; p-value (p < 0.05 is significant).

^*One-way ANOVA test.

^**Chi-Square test.

Within-group comparisons

Postural control

As shown in Table 2, the VR+taVNS group exhibited significant within-group improvements across all postural control conditions except for COP velocity in the anteroposterior (AP) direction. These improvements were reflected by reduced center of pressure (COP) displacement and velocity in both the AP and mediolateral (ML) directions, with large effect sizes (d > 0.8), indicating strong treatment effects. The VR group also showed significant reductions in AP and ML displacement under several conditions, particularly on both soft and hard surfaces with eyes open, with medium to large effect sizes. The taVNS group demonstrated selective improvements, mainly in ML displacement and AP velocity under the eyes-closed on hard surface condition, though these changes were less consistent and generally associated with small to medium effect sizes. The VR+sham-taVNS group showed minimal changes, with only a few statistically significant improvements and mostly small effect sizes. These findings suggest that the combined VR+taVNS intervention was the most effective in enhancing postural control.

Table 2.

Within-group comparisons for postural control outcome measures in different conditions (n = 7 per group).

		Eyes open on hard surface				Eyes closed on hard surface				Eyes open on soft surface				Eyes closed on soft surface
Parameter	Group	Before	After	p	d/P*	Before	After	p	d/P*	Before	After	p	d/P*	Before	After	p	d/P*
AP displacement (cm)	VR	2.49 (0.77)	1.79 (0.54)	0.018	2.41	3.75 (2.33)	3.20 (2.06)	0.018	0.98	5.16 (3.06)	4.15 (2.50)	0.018	1.50	6.48 (4.56)	5.38 (4.25)	0.028	1.41
	taVNS	2.26 (0.93)	1.86 (0.61)	0.237	0.27	3.11 (2.34)	2.80 (2.13)	0.176	0.44	5.10 (2.43)	4.96 (3.46)	0.237	0.08	6.48 (2.62)	5.63 (2.32)	0.128	0.47
	VR+sham-taVNS	2.48 (1.01)	2.15 (1.11)	0.043	0.82	3.48 (2.17)	3.14 (2.61)	0.310	0.21	4.87 (2.93)	4.35 (2.59)	0.310	0.51	6.46 (2.84)	5.77 (2.66)	0.237	0.23
	VR+taVNS	2.43 (0.83)	1.85 (0.63)	0.018	1.05	4.53 (2.04)	1.87 (0.64)	0.018	1.77	5.66 (2.51)	2.44 (1.15)	0.018	2.03	8.19 (4.19)	3.43 (0.96)	0.018	1.36
ML displacement (cm)	VR	2.33 (0.80)	1.62 (0.60)	0.018	1.57	2.78 (1.43)	2.33 (1.49)	0.128	0.36	3.45 (1.44)	3.27 (1.29)	0.043	0.85	5.42 (4.17)	4.95 (4.13)	0.128	0.49
	taVNS	1.81 (0.47)	1.38 (0.54)	0.063	0.73	2.60 (0.72)	2.04 (0.85)	0.018	1.69	3.72 (1.14)	3.02 (1.05)	0.028	0.88	5.04 (1.35)	4.89 (2.05)	0.735	0.09
	VR+sham-taVNS	1.89 (1.13)	1.53 (0.94)	0.176	0.39	2.90 (2.49)	2.45 (1.80)	0.128	0.33	3.53 (1.13)	3.18 (1.06)	0.237	0.31	5.34 (2.59)	5.10 (2.17)	0.176	0.26
	VR+taVNS	2.03 (0.54)	1.16 (0.51)	0.018	2.71	3.93 (2.90)	1.24 (0.48)	0.018	0.97	4.34 (1.35)	1.55 (0.52)	0.018	2.44	6.78 (3.11)	2.65 (1.44)	0.018	2.05
AP velocity (cm/s)	VR	0.75 (0.27)	0.68 (0.21)	0.173	0.42	1.31 (0.88)	1.16 (0.95)	0.735	0.20	1.75 (0.87)	1.45 (0.84)	0.028	1.11	2.49 (1.56)	2.22 (1.65)	0.028	0.96
	taVNS	0.64 (0.13)	0.55 (0.10)	0.128	0.35	1.01 (0.26)	0.85 (0.26)	0.018	4.00	1.20 (0.38)	1.11 (0.45)	0.176	0.18	2.34 (0.83)	2.09 (0.82)	0.028	1.04
	VR+sham-taVNS	0.70 (0.16)	0.62 (0.26)	0.292	0.28	1.03 (0.38)	0.88 (0.56)	0.176	0.52	1.42 (0.71)	1.17 (0.41)	0.176	0.20	2.50 (0.92)	1.97 (0.64)	0.043	1.03
	VR+taVNS	0.80 (0.47)	0.57 (0.21)	0.090	0.32	1.29 (0.77)	0.53 (0.15)	0.018	1.23	1.72 (0.73)	0.69 (0.20)	0.018	1.83	3.27 (1.43)	1.33 (0.92)	0.018	3.28
ML velocity (cm/s)	VR	0.64 (0.19)	0.57 (0.16)	0.310	0.24	0.97 (0.67)	0.85 (0.51)	0.128	0.39	1.32 (0.57)	1.30 (0.50)	0.735	0.06	1.96 (1.35)	2.06 (1.95)	0.735	0.09
	taVNS	0.64 (0.13)	0.51 (0.10)	0.128	0.40	0.93 (0.22)	0.73 (0.19)	0.091	0.69	1.22 (0.34)	1.00 (0.29)	0.043	0.78	2.01 (0.63)	1.91 (0.88)	0.310	0.09
	VR+sham-taVNS	0.56 (0.19)	0.49 (0.19)	0.091	0.62	0.86 (0.68)	0.81 (0.52)	0.753	0.12	1.27 (0.34)	1.01 (0.27)	0.128	0.51	2.14 (0.93)	1.87 (0.75)	0.043	0.57
	VR+taVNS	0.62 (0.16)	0.45 (0.14)	0.018	2.83	1.16 (0.73)	0.45 (0.14)	0.018	1.18	1.59 (0.45)	0.53 (0.12)	0.018	2.16	2.58 (1.31)	1.13 (0.73)	0.018	2.23

Values are presented as mean (standard deviation), the Wilcoxon signed-rank test is used, p; p-value (p < 0.05 is significant).

*Effect size (d) was calculated using Cohen’s d (small = 0.2, medium = 0.5, large = 0.8) for p < 0.05, and statistical power (P) was calculated for p > 0.05.

Vestibulo-ocular reflex gain

Because the average gains of the left and right semicircular canals were not significantly different (p > 0.05), the gains for each canal were combined for analysis. Although the VOR gain was within the normal range in all three semicircular canals (> 80%) for all groups at baseline, a statistically significant improvement was observed in all groups following the intervention (p < 0.001).

Dizziness Handicap Inventory

The mean total score and all three subscales of the DHI showed statistically significant improvement after the intervention in all groups (Table 3).

Table 3.

Within-group comparisons of Dizziness Handicap Inventory and Vertigo Symptom Scale-short form (n = 7 per group).

Questionnaire		VR				taVNS				VR+sham-taVNS				VR+taVNS
		Before	After	p	d	Before	After	p	d	Before	After	p	d	Before	After	p	d
DHI	Total	60.00 (10.95)	45.71 (11.39)	< 0.001	2.96	51.14 (22.32)	36.29 (21.64)	< 0.001	3.04	55.43 (24.48)	40.57 (19.31)	< 0.001	2.53	57.43 (22.61)	25.71 (15.07)	0.002	1.98
	Emotional	15.43 (6.99)	11.43 (4.27)	0.018	1.22	11.71 (7.95)	8.00 (6.53)	0.026	1.10	16.57 (9.28)	12.29 (7.52)	0.008	1.46	17.14 (10.25)	6.86 (5.01)	0.025	1.11
	Physical	23.43 (6.99)	17.43 (4.57)	0.022	1.16	19.71 (6.87)	12.29 (5.82)	0.006	1.57	19.71 (8.75)	14.00 (7.02)	0.001	2.12	20.29 (7.34)	10.57 (6.39)	< 0.001	2.89
	Functional	21.14 (3.43)	16.86 (5.52)	0.023	1.14	20.57 (8.54)	16.00 (10.45)	0.005	1.65	19.14 (11.18)	14.29 (8.59)	0.007	1.50	20.00 (7.48)	8.29 (7.25)	0.001	2.19
VSS-sf		26.00 (8.79)	20.57 (6.70)	0.01	1.37	21.14 (7.38)	15.14 (7.12)	< 0.001	3.00	20.14 (10.35)	13.86 (6.64)	0.006	1.59	29.00 (15.25)	8.71 (6.70)	0.002	2.04

Values are presented as mean (standard deviation), paired t-test is used, p; p-value (p < 0.05 is significant).

* Effect size (d) was calculated using Cohen’s d (small = 0.2, medium = 0.5, large = 0.8).

Vertigo Symptom Scale-short form

According to the paired t-test, the mean VSS-sf score also improved significantly in all groups (Table 3).

Frequency of vertigo/dizziness

The frequency of vertigo/dizziness decreased significantly in all groups: VR (Z = −2.06, p = 0.03, d = 0.77), taVNS (Z = −2.46, p = 0.01, d = 0.92), VR+sham-taVNS (Z = −2.42, p = 0.01, d = 0.91), and VR+taVNS (Z = −2.37, p = 0.01, d = 0.89) (Table 4).

Table 4.

Median (range) of frequency of vertigo/dizziness and headache frequency before and after intervention (n = 7 per group).

	Vertigo/dizziness		Headache
	Before	After	Before	After
VR	3 (1–6)	1 (0–5)	2 (0–6)	2 (0–6)
taVNS	4 (2–6)	2 (0–4)	3 (2–6)	1 (0–2)
VR+sham-taVNS	5 (2–6)	3 (1–4)	4 (1–6)	4 (0–5)
VR+taVNS	4 (2–6)	0 (0–2)	4 (0–6)	1 (0–2)

Frequency of vertigo/dizziness and headache was recorded and scored using a 0–6 scale: Never or less than once a week (0), once a week (1), 2–3 times a week (2), 4–6 times a week (3), once a day (4), more than once a day (5), always (6).

Frequency of headache

The frequency of headaches also decreased significantly in all groups: VR (Z = −2.00, p = 0.04, d = 0.75), taVNS (Z = −2.37, p = 0.01, d = 0.89), VR+sham-taVNS (Z = −2.00, p = 0.04, d = 0.75), and VR+taVNS (Z = −2.21, p = 0.02, d = 0.83) (Table 4).

Headache impact test

All groups showed significant improvement in mean HIT-6 scores after the intervention: VR (p = 0.002, d = 2.00), taVNS (p = 0.001, d = 2.18), VR+sham-taVNS (p < 0.001, d = 2.43), and VR+taVNS (p = 0.003, d = 1.78).

Between-group comparisons

Postural control

Postural control outcomes varied depending on the testing condition. Under the eyes-open on hard surface condition, no statistically significant differences in postural parameters were found between groups. However, in the eyes-closed on hard surface condition, significant differences were observed in the mean ranks of all postural parameters between at least one pair of groups (p < 0.05). Post hoc comparisons showed that the VR+taVNS group had significantly greater improvements than all other groups in AP displacement, greater than the VR and VR+sham-taVNS groups in ML displacement, greater than the VR and taVNS groups in AP velocity, and greater than the VR+sham-taVNS group in ML velocity. No other pairwise differences were significant. In the eyes-open on soft surface condition, significant differences in postural control measures were again observed (p < 0.05). The VR+taVNS group improved significantly more than the taVNS and VR+sham-taVNS groups in AP displacement and velocity, more than the VR and VR+sham-taVNS groups in ML displacement, and more than the VR and taVNS groups in ML velocity. In the most challenging condition—eyes closed on a soft surface—the VR+taVNS group demonstrated significantly better performance than all other groups in all parameters (p < 0.05), while no significant differences were found among the remaining groups (Figs. 1 and 2; Supplementary Table S1 online).

Fig. 1.

Comparative boxplot analysis of center of pressure displacement improvements across groups. The central line in each box represents the median; the top and bottom of each box correspond to the 75th and 25th percentiles, respectively. Whiskers indicate the minimum and maximum values, excluding outliers. AP anteroposterior; ML mediolateral; ^*p < 0.05.

Fig. 2.

Comparative boxplot analysis of center of pressure velocity improvements across groups. The central line in each box represents the median; the top and bottom of each box correspond to the 75th and 25th percentiles, respectively. Whiskers indicate the minimum and maximum values, excluding outliers. AP anteroposterior; ML mediolateral; ^*p < 0.05.

Vestibulo-ocular reflex gain

Kruskal-Wallis test revealed no statistically significant differences between groups in the mean rank improvements in VOR gain for any semicircular canal (p > 0.05).

Dizziness Handicap Inventory

Between-group comparisons revealed statistically significant differences in the improvement of the total DHI score (F(3,24) = 6.02, p = 0.003, η² = 0.43) and the functional subscale (F(3,24) = 5.92, p = 0.004, η² = 0.42), but not in the physical (F(3,24) = 1.38, p = 0.27) or emotional subscales (F(3,24) = 2.41, p = 0.09). Post hoc Bonferroni tests showed significantly greater improvements in total and functional scores in the VR+taVNS group compared to all other groups (p < 0.05), with no other pairwise differences.

Vertigo Symptom Scale-short form

A significant between-group difference in improvements in VSS-sf scores (H(3) = 11.09, p = 0.01, η² = 0.33). Post hoc comparisons showed that the VR+taVNS group improved significantly more than the VR, taVNS, and VR+sham-taVNS groups in VSS-sf scores.

Frequency of vertigo/dizziness

Significant differences in the reduction of vertigo/dizziness frequency were found between groups (H(3) = 13.72, p = 0.003, η² = 0.44). Post hoc comparisons showed greater improvements in the VR+taVNS group compared to the VR, taVNS, and VR+sham-taVNS groups (p < 0.05), with no differences between the other groups.

Frequency of headache

Significant differences were also observed in headache frequency improvements (H(3) = 15.36, p = 0.002, η² = 0.51). The VR+taVNS and taVNS groups showed significantly greater improvements than the VR and VR+sham-taVNS groups (p < 0.05), with no significant differences between other group pairs.

Headache impact test

There was a significant difference in HIT-6 score improvement between groups (F(3,24) = 7.20, p = 0.001, η² = 0.47). The VR+taVNS and taVNS groups improved significantly more than the VR and VR+sham-taVNS groups (p < 0.05). No significant difference was found between the VR+taVNS and taVNS groups.

Correlation between postural control and other outcomes

There was a moderate, positive correlation between improvements in postural control parameters under three conditions (eyes open on a soft surface, eyes closed on a soft surface, and eyes closed on a hard surface) and reductions in vertigo frequency (0.41 < rs < 0.71, p < 0.028). However, no significant correlations were found between improvements in postural control and changes in other outcome measures (p > 0.05).

Adverse events

No serious side effects were observed during the study. A few participants experienced mild, short-term skin irritation at the electrode site, which resolved spontaneously and did not cause anyone to drop out.

Discussion

This study investigated the effects of transcutaneous auricular vagus nerve stimulation (taVNS), vestibular rehabilitation (VR), and their combination on postural control, dizziness, and headache in patients with vestibular migraine (VM). Four intervention groups were compared to assess the individual and combined benefits of these approaches. The results are discussed with reference to both objective balance measures and subjective reports of dizziness and headache.

Postural instability, imbalance, and intolerance to head movement are among the most disabling symptoms in patients with VM^24,25. Our findings demonstrated improvements in specific aspects of postural control across all groups, with effect sizes exceeding 0.8, indicating strong and clinically meaningful intervention effects²⁶. Notably, the VR+taVNS group showed significantly greater improvements than the other groups under three testing conditions: eyes closed on a hard surface, eyes open on a soft surface, and eyes closed on a soft surface.

Vestibular rehabilitation targets the vestibular system to enhance central nervous system plasticity, restore balance, and promote compensation, adaptation, and habituation. Its core strategies include desensitization of the vestibular system, strengthening of vestibulo-ocular and vestibulo-spinal reflexes, and developing alternative sensory strategies to address balance deficits and reduce dizziness and vertigo²⁷. Significant improvements have been reported in DHI scores, balance confidence, gait, and posture in patients with migraine-related dizziness following targeted VR exercises²⁸. Similarly, VR improved posturographic measures in both VM and non-migraine patients, supporting its broad utility²⁹. In our study, participants in the VR group also showed significant improvements in specific postural control parameters, indicating enhanced vestibular function.

The beneficial effects of taVNS on VM may be attributed to the interconnections of the vestibulo-vagal system³⁰. Although taVNS is applied peripherally at the cymba and cavum conchae, its therapeutic effects are primarily mediated through central mechanisms via afferent vagal projections to the nucleus tractus solitarius and subsequent modulation of widespread brain regions involved in vestibular processing, autonomic regulation, and pain control^31,32. Therefore, taVNS can be considered a peripheral neuromodulatory intervention that exerts its effects predominantly through central vestibular and autonomic networks³¹.This central integration and the interconnected pathways between the vagal, vestibular, and trigeminal systems are schematically illustrated in Fig. 4C. Vagal afferents and efferents terminate in four medullary nuclei: NTS, nucleus ambiguus, spinal trigeminal nucleus, and the dorsal motor nucleus of the vagus³³. The NTS, as the primary relay for vagal afferents, contains neurocircuitry that links the trigeminal, vestibular, and vagal systems and is implicated in motion sickness and nausea associated with migraines^30,34. Efferent projections from vagal nuclei to the vestibular system are important for motor control during upright stance. There are also established interactions between the vagal and vestibular systems³⁵, with anterograde and retrograde tracing studies showing projections from the caudal medial and inferior vestibular nuclei to both the NTS and the dorsal motor nucleus of the vagus³⁶.

Fig. 4.

Transcutaneous auricular vagus nerve stimulation (taVNS) procedure and proposed central mechanism in vestibular migraine. (A) Real taVNS with electrode placement on the left cymba conchae and cavum conchae. (B) Sham taVNS with electrode placement on the tail of the helix. Photographs taken during the study with patients’ consent. (C) Schematic diagram of the central pathway: stimulation of the auricular branch of the vagus nerve (ABVN) projects afferents to the nucleus tractus solitarius (NTS), activating the locus coeruleus, raphe nuclei, and parabrachial nuclei, leading to modulation of vestibular nuclei and the trigeminovascular system.

To date, no prior studies have examined the effect of taVNS on postural control in VM. While the exact mechanism remains unclear, improvements can be attributed to taVNS’s analgesic and anti-inflammatory effects, as well as its impact on psychological well-being, such as reducing depression and improving mood^37,38. The vestibular and autonomic systems are anatomically linked via vestibular-autonomic reflexes^39,40, which contribute to cardiovascular and respiratory regulation. Vagus nerve stimulation has also been shown to modulate activity in brain areas including the insula and cingulate cortex³¹, both of which are known to play critical roles in central postural control⁴¹. These shared neural networks may explain the positive impact of taVNS on balance. For example, in patients with PPPD, non-invasive vagus nerve stimulation significantly improved quality of life, reduced depression, and decreased postural sway as measured by posturography¹⁹. In line with this, VM patients in our taVNS group also exhibited improvements in specific postural control parameters.

Our results further demonstrated that participants in the VR+taVNS group experienced greater improvements in postural control than those in the other groups. Since combination therapy may result in synergistic effects, the concurrent application of VR and taVNS likely yielded enhanced outcomes. Supporting this, a study involving patients with chronic lower back pain found that adding taVNS to exercise therapy significantly improved dynamic balance⁴².

To subjectively assess the effects of VR and taVNS on vertigo, dizziness, and headache in VM patients, we employed the DHI, VSS-sf, and HIT-6, along with measures of vertigo/dizziness and headache frequency. All groups demonstrated statistically significant improvements in these symptoms. The marked reductions in VSS-sf and DHI total and subscale scores across all groups indicate that both VR and taVNS are effective in reducing balance impairments, vestibular symptoms, and functional disability. These results align with previous studies investigating cervical nVNS in PPPD¹⁹ and Meniere’s disease²¹, as well as VR in VM^43,44 and migraine-related dizziness⁴⁵. Consistent with our findings, Shaabani et al. also reported decreased dizziness frequency in VM patients following VR⁴³.

In the present study, the VR+taVNS group exhibited significantly greater improvement in the DHI total and functional subscale scores, the VSS-sf scores, and the frequency of vertigo/dizziness compared to the other groups. These findings highlight the enhanced efficacy of the combined VR and taVNS intervention in mitigating the severity and frequency of vestibular symptoms in VM.

Regarding headache outcomes, assessed via HIT-6 and headache frequency, all groups showed statistically significant improvements. However, participants in the VR+taVNS and taVNS groups experienced greater improvements than those in the VR and VR+sham-taVNS groups. Sugaya et al. previously reported positive effects of VR on headache severity in migraine patients, though their study lacked a control group¹². Similarly, Shaabani et al. found improvements in both HIT-6 scores and headache frequency following VR in VM patients⁴³.

Prior studies have also demonstrated beneficial effects of nVNS in migraine treatment^46,47. For instance, Najib et al. reported that taVNS improved HIT-6 scores and supported its clinical utility in migraine prevention⁴⁷. Another study found that both 25 Hz and 1 Hz auricular taVNS improved HIT-6 scores in chronic migraine patients, with 1 Hz stimulation showing greater efficacy⁴⁸. Potential mechanisms of vagal stimulation include inhibition of CNS excitability by reducing glutamate release, suppression of trigeminocervical nociceptive activation, and downregulation of central sensitization-related proteins^49,50. The analgesic effects of low-frequency electrical stimulation have been demonstrated in both human and animal pain models⁵¹, suggesting its potential as a pain management strategy. Although the exact pathway remains unknown, it is hypothesized that vagal stimulation may inhibit nociceptive trigeminal neurons either directly or indirectly, possibly through activation of descending pain inhibition systems via NTS projections to the nucleus raphe magnus and locus coeruleus—key sites in serotonergic and noradrenergic pain modulation⁵².

Finally, this study revealed a moderate, positive correlation between improvements in postural control and reductions in vertigo/dizziness frequency. Vertigo involves a distributed network encompassing the posterior insular vestibular cortex, somatosensory and visual cortices, the cingulate sulcus visual area, and the thalamus⁵³. The vestibular cortex works in concert with the somatosensory and visual systems to control posture, gait, and gaze. As the insula and cingulate cortices are also integral to postural regulation⁴¹, the observed improvements in subjective symptoms may reflect the impact of VR and taVNS on shared neural substrates involved in both balance and dizziness perception¹⁷. A similar approach using bimodal stimulation has also shown promising results in alleviating auditory-related symptoms such as tinnitus, supporting the neuromodulatory effects of taVNS⁵⁴.

This study is the first to assess the combined effects of VR and repeated sessions of auricular taVNS in VM patients using both objective and subjective measures. The use of multiple validated tools enhances the comprehensiveness of the findings. Given the significant clinical benefit and minimal, transient skin irritation, the combined intervention has a highly favorable risk-benefit profile for patients with VM.

However, limitations include a small sample size that reduced statistical power, the absence of a post-treatment washout period, a predominantly female sample, which may limit generalizability, and the lack of complete blinding for both participants and the administering audiologist due to the nature of interventions, which could cause some performance and expectation bias. Although the statistician analyst was kept blinded to reduce this risk. Additionally, the absence of a pure no-intervention control group (e.g., waiting list or conventional treatment) is a limitation. All participants received some form of active treatment, which precludes fully distinguishing the specific effects of the interventions from potential natural disease course, regression to the mean, or non-specific therapeutic effects. Another limitation is the absence of formal vestibular migraine severity grading at baseline. Although symptom severity was assessed using validated subjective and objective measures, patients were not classified by standardized VM severity classifications. The findings of this study are applicable to adults (18–50 years) with confirmed vestibular migraine (ICHD-3 criteria).

Future studies should involve larger, more balanced samples and include long-term follow-up to assess the durability of effects. Severity-based subgroup analyses are recommended to determine whether treatment efficacy varies with vestibular migraine severity. However, because we ran this trial in a specialized clinic, using trained staff and specific equipment, it may be hard to apply these results in general health clinics or places with fewer resources. To be sure about these findings, we need more studies in different real-world settings with various patient groups. Furthermore, the safety and effectiveness of taVNS and vestibular rehabilitation should be evaluated in pediatric patients.

Conclusion

This study demonstrates that both VR and taVNS independently improve postural control, dizziness, and headache symptoms in patients with vestibular migraine. Importantly, the combined application of VR and taVNS produced greater and clinically meaningful improvements in both objective balance measures and subjective reports of dizziness, vertigo, and headache. The observed synergistic effect may result from complementary mechanisms, including enhanced CNS plasticity, vestibular-autonomic integration, and modulation of pain and sensory pathways. These findings support the combined use of VR and taVNS as a promising, noninvasive therapeutic approach for managing the multifaceted symptoms of VM. Future research involving larger samples and extended follow-up is warranted to confirm these effects and elucidate the underlying mechanisms.

Methods

Study design

This study was a parallel-group, randomized controlled trial with a 1:1:1:1 allocation ratio, comparing four groups: VR, taVNS, VR+sham-taVNS, and VR+taVNS. The study was designed to evaluate the superiority of VR+taVNS in improving dizziness and postural control over one month. The trial was approved by the Ethics Committee of Tehran University of Medical Sciences (Code: IR.TUMS.FNM.REC.1401.160), and all methods were performed in accordance with the relevant guidelines and regulations. This study is registered with the Iranian Registry of Clinical Trials (No: IRCT20160131026279N5) on 25/06/2023. The full trial protocol is available for public access via this registry. Participants were recruited between February 23, 2023, and May 21, 2024. The final assessment was completed on June 22, 2024.

Participants

Participants were patients with vestibular migraine diagnosed according to the ICHD-3 criteria⁵ by a single board-certified neurologist/headache specialist with extensive clinical experience in headache disorders. After completing a course of medical therapy, they were referred from the Headache Tertiary Center at Sina Hospital, Tehran University of Medical Sciences. All participants had completed a stable course of pharmacological treatment by the study neurologist and had no changes in prophylactic medication for at least two months before enrollment and throughout the study period. All interventions and data collection were conducted at the Vestibular Assessment and Rehabilitation Clinic, School of Rehabilitation, Tehran University of Medical Sciences, Tehran, Iran. All assessments and interventions were performed by an audiologist.

Inclusion criteria were: age between 18 and 50 years; no changes in prophylactic medication during the past two months or during the study; no use of vestibular-suppressant medications; no other vestibular or neurological disorders; no visual impairments; no orthopedic problems in the past six months; absence of lower limb deformities, rheumatic or metabolic diseases, neck pain, or limitations in neck range of motion; and no history of alcohol or sedative use. Patients with prior vestibular rehabilitation were excluded.

Before enrollment, all participants underwent a comprehensive otoneurological evaluation to confirm the diagnosis of vestibular migraine and exclude other vestibular disorders. This evaluation included detailed clinical history, neurological examination by the headache specialist, detailed medical and vestibular history taking, bedside otoneurological examination, and videonystagmography (VNG).

The sample size was calculated using G*Power (version 3.1). As no prior study with the same design in vestibular migraine was available, effect sizes were derived from a previous four-arm study⁵⁵ that examined the effects of combined vestibular rehabilitation and electrical stimulation on postural control in patients with vestibular dysfunction. Between-group effect sizes for sway displacement and velocity (primary outcomes) during standing on foam with eyes open ranged from η² = 0.31 to 0.56. Assuming 80% power and α = 0.05, the required total sample size was estimated to be 14–28 participants. Accordingly, 28 participants (7 per group) were included in the final analysis. A total of 34 patients were initially recruited, of whom 28 (22 females; mean age = 40.39 ± 6.29 years) completed the training sessions (Fig. 3). All participants provided written informed consent.

Fig. 3.

Flowchart of the study. DHI dizziness handicap inventory; VSS-sf Vertigo Symptom Scale-short form; HIT-6 headache impact test; vHIT video head impulse test.

Participants were monitored for one month during a run-in phase to record headaches, vertigo attacks, and associated symptoms. They were then randomly allocated into four intervention groups. The random allocation sequence was generated by an independent researcher using the Random Allocation Software. Block randomization with a block size of 4 was applied to maintain balance among the groups throughout the enrolment period. For allocation concealment, the sequence was secured using sequentially numbered and sealed envelopes, which were opened by the researcher only after eligibility was confirmed and informed consent was obtained. The VR group received one month of vestibular rehabilitation consisting of daily home-based exercises targeting balance and gaze stability. The taVNS group underwent transcutaneous auricular vagus nerve stimulation delivered in twelve 30-minute sessions over the course of one month, administered three times per week. The VR+sham-taVNS group received the same vestibular rehabilitation program combined with twelve 30-minute sessions of sham taVNS. The VR+taVNS group received a combination of vestibular rehabilitation and real taVNS. Because of differences in treatment protocols (i.e., presence or absence of VR and taVNS), participants and the administering audiologist could not be completely blinded to the groups, which may introduce potential performance and expectation biases. To minimize bias, the real and sham taVNS used the same device and electrodes, but the sham delivered no stimulation. Furthermore, to ensure the objectivity of the findings, all data were analyzed by a statistical analyst blinded to group assignments.

Outcome measures

At baseline (on the same day and prior to the first intervention session) and at the end of the study (after the final intervention session on day 30), participants were assessed for postural control parameters as primary outcomes, and video head impulse test (vHIT), the Persian versions of the Dizziness Handicap Inventory (DHI) and Vertigo Symptom Scale–short form (VSS-sf), frequency of vertigo/dizziness, frequency of headache, and the Headache Impact Test (HIT-6) as secondary outcomes. All assessments were performed in a randomized order to prevent order effects.

Postural control assessment

Postural control was assessed using a force platform (Bartech 9090–15; Bartech Corp., Columbus, OH), which recorded center of pressure (COP) trajectory data at 500 Hz. The X and Y axes represented mediolateral (ML) and anteroposterior (AP) COP displacements, respectively. Participants stood barefoot with their feet 5 cm apart under four conditions: eyes open on hard surface, eyes closed on hard surface, eyes open on soft surface, and eyes closed on soft surface. A 10 cm-thick foam pad was placed on the platform to simulate the soft surface.

During each trial, participants were instructed to breathe normally, avoid deep breathing, yawning, or coughing, and keep their arms relaxed at their sides while focusing on a fixed point four meters ahead. Each condition was repeated three times with short breaks between trials. Each trial lasted 35 s, and data were sampled at 100 Hz.

COP data were processed in MATLAB (version 7.7.0471) using a zero-lag, second-order Butterworth low-pass filter (cut-off frequency: 4 Hz). The first 5 s were excluded, and the remaining 30 s were used for analysis. Outcomes included COP displacement and velocity in both AP and ML directions. Displacement was defined as the range (maximum–minimum) of COP excursion in each direction, and velocity was calculated as total excursion divided by time. The mean of three trials per condition was used for statistical analysis^56,57.

Video head impulse test (vHIT)

A video-based system (vHIT, GN Otometrics, Denmark) was used to measure the vestibulo-ocular reflex (VOR) in each semicircular canal. Approximately twenty quick, unpredictable head impulses (10–20°) were administered while the patient focused on a fixed target. These were delivered in the yaw axis to test the horizontal canals. For LARP (left anterior/right posterior) and RALP (right anterior/left posterior) canal evaluation, the patient’s head was rotated 35–45° to the right or left, respectively, and head impulses were delivered in the pitch axis. VOR gain values < 0.8 (horizontal canals) and < 0.7 (vertical canals) were considered indicative of deficit⁵⁸.

Dizziness Handicap Inventory (DHI)

The Persian version of the DHI was used to assess perceived disability in patients with dizziness⁵⁹. The questionnaire contains 25 items across three subscales (physical, functional, and emotional), with a total score ranging from 0 to 100.

Vertigo Symptom Scale-short form (VSS-sf)

The Persian version of the VSS-sf was used, consisting of 15 items scored on a 5-point scale (0–4) to evaluate dizziness symptoms⁶⁰. The total score ranges from 0 to 60, with higher scores indicating greater symptom severity. A score ≥ 12 reflects severe dizziness.

Frequency of vertigo/dizziness

Vertigo/dizziness frequency over one month was recorded and scored using a 0–6 scale: Never or less than once a week (0), once a week (1), 2–3 times a week (2), 4–6 times a week (3), once a day (4), more than once a day (5), always (6)⁴³.

Frequency of headache

Headache frequency was rated using the same 0–6 scale as vertigo/dizziness frequency⁴³.

Headache impact test (HIT-6)

The Persian version of HIT-6 was used to assess the impact of headache on functioning⁶¹. It contains six items addressing pain severity, role functioning, social functioning, vitality, emotional distress, and cognitive functioning. Responses range from “never” to “always,” with scores from 6 to 13 per item. The total score ranges from 36 to 78, with higher scores indicating greater impact.

Interventions

Vestibular rehabilitation (VR)

Patients received a one-month VR program based on a protocol used in previous migraine studies¹². The program included 20 exercises: seven targeting the vestibulo-ocular reflex (VOR) (e.g., eye tracking and head movements with fixed gaze), eight static vestibulo-spinal reflex (VSR) exercises (e.g., tandem and single-leg stance), and five dynamic VSR exercises (e.g., walking with head turns). Each daily session lasted approximately 20–30 min. All exercises were taught in the first session, and each patient received a checklist and an illustrated booklet. They were instructed to complete the exercises daily and log them in the checklist. Follow-up calls were made twice weekly to ensure compliance, and the completed checklists were reviewed at the end of treatment. To ensure the validity of the results, a minimum adherence threshold of 80% (completion of exercises on at least 24 out of 30 days) was set. Participants who did not meet this criterion were excluded from the final analysis, as shown in the study flowchart.

Patients diagnosed with both VM and benign paroxysmal positional vertigo first received repositioning maneuvers before beginning VR.

Transcutaneous auricular vagus nerve stimulation (taVNS)

taVNS was administered using the Neurostim2 electrical current generator (Medina Teb Co., Iran) and ear-clip electrodes. Skintact ECG electrode gel (Leonhard Lang, Austria) was used to ensure good electrode contact. The protocol involved 12 morning sessions over 4 weeks, with continuous 30 min per session. Stimulation was applied at 1 Hz with 0.2 ms pulse duration. During stimulation sessions, participants were seated comfortably. The intensity was set at the highest tolerable level without pain (1.5–5 mA). In the taVNS and VR+taVNS groups, stimulation targeted the left cymba and cavum conchae, which are the auricular regions with the highest density of vagal afferents and have been shown to project centrally to the nucleus tractus solitarius (NTS) and associated brain regions involved in vestibular processing, autonomic regulation, and pain modulation^31,32,62. Proper electrode placement was ensured visually by the administering audiologist, and skin impedance was checked before each session to confirm adequate contact and stimulation.

In the VR+sham-taVNS group, the device was placed on the tail of the helix, turned on for the same application time as the real intervention, but no current was delivered, ensuring no physiological stimulation⁶². Electrode placement for real and sham taVNS is shown in Fig. 4A and B.

Statistical analysis

All statistical analyses were performed using IBM SPSS Statistics (Version 17; IBM Corp., Armonk, NY, USA). Tests were two-tailed, and statistical significance was defined as p < 0.05. Descriptive data are presented as mean ± standard deviation (SD) for continuous variables and as median (range) for discrete variables (e.g., frequency of vertigo/dizziness and headache). The Kolmogorov–Smirnov test was applied to assess the normality of distributions, and Levene’s test was used to evaluate the homogeneity of variances.

Sex distribution across groups was compared using the chi-squared test. Between-group comparisons (n = 7 per group) for baseline characteristics and intervention-related changes (post–pre differences) were conducted using one-way analysis of variance (ANOVA) for normally distributed data or the Kruskal–Wallis test for non-normally distributed data, with Bonferroni correction applied to post hoc analyses to control for multiple comparisons and reduce the risk of type I error. Within-group changes were examined using paired t-tests for normal data or Wilcoxon signed-rank tests for non-normal data.

Postural control parameters were defined as the primary outcomes. Other variables, including DHI and VSS-sf scores, frequency of vertigo/dizziness, frequency of headache, and the HIT-6 score, were considered secondary outcomes.

Effect sizes were estimated using Cohen’s d (small = 0.2, medium = 0.5, large ≥ 0.8)²⁶, while between-group effect magnitudes were determined using eta squared (η²; small = 0.01, medium = 0.06, large = 0.14)⁶³. Correlations between improvements in postural control and other outcome measures were assessed using Spearman’s rank-order correlation coefficient.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

AF: Acquisition of data, drafting the manuscript, analysis, and interpretation of data; MAG: Study concept and design, study supervision, drafting/critical revision of the manuscript for important intellectual content; MT: Selection and evaluation of patients for recruitment and follow-up, study supervision, and neurological editing of the manuscript; KM: technical and Material support, and assistance with manuscript editing; SJ: Study design, analysis, and interpretation of data; All authors read and approved the manuscript.

Data availability

The datasets used and/or analyzed during the present study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.