Effects of vibration stimulation on truncal lateropulsion in patients with lateral medullary syndrome Authors: Masanori Nagaoka, Shoma Shiokawa, Yasuhiro Kumakura, Tomo Iwasaru, Yasuhiro Utsugi, Natsuki Shoda, Jiro Fukae, Natsuko Yamamoto, Yumi Chiba Published: 2026 ABSTRACT AbstractBackgroundPrognosis of body lateropulsion observed in lateral medullary syndrome is generally favourable; however, lateropulsion persists in some patients and delays social rehabilitation. One proposed pathophysiological mechanism of lateropulsion is impaired afferent input to the cerebellum due to disruption of the posterior spinocerebellar tract. This study examined whether activation of the posterior spinocerebellar tract system through vibration stimulation improves lateropulsion.MethodsFirst, following a representative case presentation, seven additional consecutive patients with lateral medullary syndrome were studied to analyse the immediate effects of vibration stimulation evaluated by the Romberg test, tandem standing test and 3 m Timed-Up-and-Go test. Next, a prospective observational study of vibration stimulation on balance disorders was performed in three patients who participated in a combined training approach (hybrid training) incorporating vibration stimulation and conventional balance training.ResultsBalance function before vibration stimulation was assessed in seven additional patients using three tests. Regarding the immediate effect of vibration stimulation, all patients showed improvement in at least one test when vibration was applied to the lower limb on the lesional side, whereas only three patients improved with vibration on the non-lesional side. This difference indicated significant effectiveness of lesional-side stimulation (p<0.05, Fisher’s exact test). There were no tests with worse scores after vibration. Long-term effects of hybrid training were evaluated by monitoring balance tests. Due to the limitations of a single-case observational study, it was not possible to objectively establish superiority of the hybrid training over conventional balance training.ConclusionsWe showed that vibration stimulation on lesional side is potentially useful to improve spinocerebellar dysfunction and lateropulsion. WHAT IS ALREADY KNOWN ON THIS TOPIC Trunk lateropulsion is a cardinal symptom of lateral medullary syndrome, putatively caused by impaired transmission of unconscious proprioceptive information from the lower trunk and limbs to the cerebellum, due to a unilateral lesion in the spinocerebellar tract. It is believed that imbalance of proprioception is gradually compensated over time leading to improvement of lateropulsion. Trunk lateropulsion is a cardinal symptom of lateral medullary syndrome, putatively caused by impaired transmission of unconscious proprioceptive information from the lower trunk and limbs to the cerebellum, due to a unilateral lesion in the spinocerebellar tract. It is believed that imbalance of proprioception is gradually compensated over time leading to improvement of lateropulsion. WHAT THIS STUDY ADDS 10 min of vibration stimulation applied to the lower limb on the lesional side improved at least one of the Romberg test, tandem standing test and 10 m Timed-Up-and-Go test, whereas stimulation applied to the non-lesional side showed no effect, suggesting that vibration stimulation improves proprioception imbalance caused by spinocerebellar tract lesions. 10 min of vibration stimulation applied to the lower limb on the lesional side improved at least one of the Romberg test, tandem standing test and 10 m Timed-Up-and-Go test, whereas stimulation applied to the non-lesional side showed no effect, suggesting that vibration stimulation improves proprioception imbalance caused by spinocerebellar tract lesions. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Posterior and anterior spinocerebellar tracts have been widely studied in animals, but their functions remain unclear. Use of vibration stimulation may contribute to advance future research on balance controlling mechanisms including spinocerebellar tract. Posterior and anterior spinocerebellar tracts have been widely studied in animals, but their functions remain unclear. Use of vibration stimulation may contribute to advance future research on balance controlling mechanisms including spinocerebellar tract. Introduction Lateral medullary syndrome, also known as Wallenberg syndrome, is caused by circulatory disturbances in the vertebrobasilar arterial system. Depending on the extent of the ischaemic lesion, symptoms may include ataxia of ipsilateral upper and lower limbs, lateropulsion (truncal deviation to one side during gait, without weakness (hemiparesis) and caused by medullary lesion. Symptoms may include Horner’s syndrome and dissociated sensory disturbances (impaired pain and temperature sensation on ipsilateral face as well as contralateral trunk and limbs).1 Dysphagia and, in rare cases, neurogenic bladder have also been reported.2 Lateropulsion is often transient and resolves spontaneously within several days to weeks, but in some cases the condition persists, which may delay social rehabilitation. Factors potentially involved in gait disturbances in lateral medullary syndrome include the vestibulospinal tract affecting muscle tone3; vestibular nucleus associated with nystagmus, vertigo and axial inclinations4; lesions in the spinocerebellar tract that transmits unconscious proprioceptive information to the cerebellum for motor control46; and medial lemniscus lesions that impair contralateral passive kinesthesia and sense of discrimination.4 7 The cerebellar hemisphere, which integrates the sensory inputs of the above mechanisms, has been implicated in body lateropulsion.8 9 Under normal conditions, unconscious proprioceptive input from the lower limbs/trunk is transmitted to the cerebellum via the spinocerebellar tract and contributes to motor control.4 In lateral medullary syndrome (Wallenberg syndrome), ischaemic lesion in the lateral medulla may disrupt unilateral anterior or posterior spinocerebellar tract, interrupting the information transmission from the lower limb to the cerebellum and potentially causing gait lateropulsion. Previous studies have attempted to substitute unconscious proprioception with somatosensory input to facilitate motor learning.1014 In our previous report of a patient diagnosed with lateral medullary syndrome admitted to a rehabilitation hospital in Japan, with a convalescent rehabilitation ward providing high volume training 3 hours every day during the hospitalisation period,15 we applied vibration stimulation to the lower limb of the lesional side aiming to enhance proprioceptive input.16 The stimulation resulted in remarkable improvement in posture and gait disturbance. In the present study, we subsequently administered vibration stimulation to seven consecutive patients with the same condition and observed improvements in balance function. This report proposes the potential of vibration stimulation as a novel therapeutic approach for gait disturbance in lateral medullary syndrome. Methods Subjects Between July 2023 and July 2025, eight consecutive patients diagnosed with lateral medullary syndrome were admitted to our subacute rehabilitation hospital (table 1). Table 1Clinical profile in individual patientsCaseAgeSexTime between onset andTime between admission andTime between vibration stimulation andIpsilesionalContra-lesionalLesion sideAdmission (days)Vibration stimulation (days)Discharge (days)Independent gait (days)Limb ataxiaSensory disturbance of faceNystagmus/vertigo/vomitHorner syndromeDysphagia/hoarsenessLateropulsionToBLS- walkingSensory disturbance of trunk and limbs180sM25116349+++?++Rt3+Rt260sM193241±――+―+Rt2―Rt330sM1739136-7+―++++Rt2+Rt480sM232193−18++++++Rt1+Rt540sF3911105−1++++―+Rt2+Rt660sM30188653++++++Lt3+Lt750sM2002929++++―+Lt3+Lt850sM1614478++++―+Rt2+RtMean61.524.014.672.926.4SD17.48.012.140.621.9BLS, Burke Lateropulsion Scale; Lt, left; Rt, right. Clinical profile in individual patients BLS, Burke Lateropulsion Scale; Lt, left; Rt, right. Case 1, a man in his 80s diagnosed with right lateral medullary syndrome following atherothrombotic cerebral infarction (figure 1) was transferred to our hospital for rehabilitation (see details of signs and symptoms in online supplemental text). Balance training was initiated, although severe impairment prevented unsupported standing or walking (Burke Lateropulsion Scale (BLS))-walking score 317) (see online supplemental video). To explore whether direct activation of muscle spindle activity in the lower limb—an important input source for the spinocerebellar tract—improves balance disorder, we administered vibration stimulation to the lower limb on the lesional side. This treatment improved walking speed, narrowed step width, and reduced balance impairment which were judged on the same day before and after the vibration stimulation (table 2; see online supplemental video). Case 1: Head MRI. (A) Diffusion-weighted MRI shows hyperintensity at the lateral aspect of right (Rt) medulla oblongata. (B) Transverse sections through the medulla oblongata caudal to the inferior cerebellar peduncles. Hyperintense areas, in which posterior spinocerebellar tract might be involved, are located in the lateral part of the lower portion of the medulla oblongata on the Rt side. At these levels, vestibulospinal tract, which is also supposed to be the lesion causing lateropulsion, should be located ventrally. Lt, left. Table 2Case 1: effects of vibration stimulation on gait impairmentNo.ConditionWalk distanceTime taken (gait speed)ResultFirst trial 1Before stimulation (baseline)10 m46 s (13 m/min)Baseline 2After stimulation of right leg10 m32 s (18 m/min)40% increase vs baselineSecond trial (3 days after the first trial) 3After stimulation of left leg8 m38 s (12.6 m/min)No remarkable effect vs baseline 4After stimulation of right leg8 m25 s (19.2 m/min)Marked effect vs baselineIn the first trial, 1 day after conventional gait training, the patient (case 1) was asked to walk as fast as possible for a 10 m distance (1). Then, vibration stimulation was applied to the right leg for 15 min, and he performed the 10 m walk again (2). The time taken for the 10 m was recorded. Gait speed increased by 40% after vibration stimulation compared with before stimulation (baseline). Step width decreased from 30 cm before stimulation to 15 cm after stimulation. In the second trial, conducted 3 days after the first trial, the effect of vibrational stimulation of the left (contralesional) leg was first examined (3), showing no improvement in gait speed compared with baseline. On the same day, the effect of vibration stimulation of the right (ipsilesional) leg was reassessed (4), which showed marked improvement compared with baseline. Case 1: effects of vibration stimulation on gait impairment In the first trial, 1 day after conventional gait training, the patient (case 1) was asked to walk as fast as possible for a 10 m distance (1). Then, vibration stimulation was applied to the right leg for 15 min, and he performed the 10 m walk again (2). The time taken for the 10 m was recorded. Gait speed increased by 40% after vibration stimulation compared with before stimulation (baseline). Step width decreased from 30 cm before stimulation to 15 cm after stimulation. In the second trial, conducted 3 days after the first trial, the effect of vibrational stimulation of the left (contralesional) leg was first examined (3), showing no improvement in gait speed compared with baseline. On the same day, the effect of vibration stimulation of the right (ipsilesional) leg was reassessed (4), which showed marked improvement compared with baseline. Based on this finding, we further investigated whether vibration stimulation improves balance function and gait lateropulsion in seven subsequent patients with lateral medullary syndrome admitted between December 2024 and July 2025. Clinical profiles of the eight patients are shown in table 1. Case 1 consented to publication of medical information and video. Vibration stimulation was performed as in case 1. A commercially available massager (MD-001S, Daito Denki Kougyou, Osaka) was used at a frequency of approximately 110 Hz. Stimulation was administered to the muscles of the lower limb on the lesional side (ipsilateral to the lesion) and occasionally to the buttock and trunk muscles. The stimulation was typically administered for 10 to 15 min, with changes in vibration site. For comparison, stimulation was also applied to the non-lesional side (contralateral to the lesion). Gait lateropulsion was assessed according to BLS-walking score.17 The effect of vibration stimulation was evaluated by conducting three balance tests and recording with a video camera before and after the vibration stimulation. The tests were the Romberg test, tandem standing test (Mann posture), and 3 m Timed-Up-and-Go (TUG) test18 (table 3). Romberg test and tandem test were conducted according to the methods of the Berg Balance Scale (BBS).19 The criteria of judgement were described in the footnote of table 3. The TUG test was performed according to Podsiadlo and Richardson.18 Additionally, subjects were asked to describe their experience of the vibration stimulation. In two patients (cases 1 and 6) who showed effects, the Romberg or tandem test was repeatedly conducted after a single 10-minute stimulation on the lesional side to examine how long the effect lasted. The recorded video footage was evaluated by rehabilitation staff not in charge of the subjects. Evaluation criteria were based on the items in BBS: the Romberg test corresponded to the item ‘standing unsupported with eyes closed’, and the tandem standing test to the item ‘standing unsupported one foot in front’ (table 3). The immediate effect of vibration stimulation was considered to be present if there was an increase in score of 1 or more on the BBS scale for the Romberg or tandem test, or an improvement in speed of 10% or more for the TUG test. Table 3Effects of vibration stimulation in individual patientsCaseAgeSexAdmission BLS- walking scoreLesion sideTestLesional side stimulationContralesional side stimulationOrder of stimulation sideInterval of two testsChange of BLS-walking scoresIntrospectionPrePostPrePostPrePost180sM3RtGait speed13 m/min18 m/min12.6 m/min*L→Cont3 days later21I feel easier to stand up and walk with good posture19.2 m/min† 260sM2RtRomberg1444L→Cont3 days later11(After tandem test) I feel more stableTandem0404‡TUG10108.18.2330sM2RtRomberg4444L→Contsame day11(After tandem test) I feel my balance has improvedTandem04‡00TUG14141414480 sM1RtRomberg4444L→Contsame day00I feel I can walk more steadilyTandem4444TUG8.87.78.87.4540sF2RtRomberg44§44L→Contsame day11I can now lift my right arm properlyTandem0404‡TUG10.510.610.510.6660sM3LtRomberg2444L→Cont4 days later22I feel my left leg becomes lighter and easier to moveTandem0000TUG––––750sM3LtRomberg0400Cont→Lsame day33I feel I can put more strength into my left legTandem0000TUG––––850sM2RtRomberg4444L→Cont4 days later21I feel it’s effectiveTandem0444TUG10.47.87.47Yellow-filled cell shows positive immediate effect. Romberg test: Standing unsupported with eyes closed. Score 4, able to stand 10 s safely; 3, able to stand 10 s with supervision; 2, able to stand 3 s; 1, unable to keep eyes closed 3 s but stay safely; 0, needs help to keep from falling. Increase in score of 1 or more was judged as effective.Tandem test: Standing unsupported one foot in front. Score 4, able to place foot in tandem independently and hold 15 s; 3, able to place foot ahead independently and hold 15 s; 2, able to take small step independently and hold 15 s; 1, needs help to take step but can hold 15 s; 0, loses balance while stepping or standing. BLS walking score: Score 0, no lateropulsion noted; (1) mild lateropulsion noted (no walking aid needed); (2) moderate lateropulsion noted (walking with walker) and (3) strong lateropulsion noted (needs assistance from two persons to walk or unable to walk because of severe lateropulsion, use wheel chair).*Contralesional side vibration stimulation done 3 days after lesional side stimulation.†Lesional side stimulation done on same day after contralesional side stimulation; 'elapsed time was shorter than the criteria.‡Elapsed time was shorter than the criteria.§Posture related to hypotonus improved after vibration stimulation; Romberg, Tandem: increase in score ≥was judged as effective; TUG: >10% decrease in elapsed time (10% faster) was judged as effective-, Impossible to do; BLS, Burke Lateropulsion Scale; Cont, contralesional; Lt, left; Rt, right; TUG, Timed-Up-and-Go. Effects of vibration stimulation in individual patients Yellow-filled cell shows positive immediate effect. Romberg test: Standing unsupported with eyes closed. Score 4, able to stand 10 s safely; 3, able to stand 10 s with supervision; 2, able to stand 3 s; 1, unable to keep eyes closed 3 s but stay safely; 0, needs help to keep from falling. Increase in score of 1 or more was judged as effective. Tandem test: Standing unsupported one foot in front. Score 4, able to place foot in tandem independently and hold 15 s; 3, able to place foot ahead independently and hold 15 s; 2, able to take small step independently and hold 15 s; 1, needs help to take step but can hold 15 s; 0, loses balance while stepping or standing. BLS walking score: Score 0, no lateropulsion noted; (1) mild lateropulsion noted (no walking aid needed); (2) moderate lateropulsion noted (walking with walker) and (3) strong lateropulsion noted (needs assistance from two persons to walk or unable to walk because of severe lateropulsion, use wheel chair). Contralesional side vibration stimulation done 3 days after lesional side stimulation. Lesional side stimulation done on same day after contralesional side stimulation; 'elapsed time was shorter than the criteria. Elapsed time was shorter than the criteria. Posture related to hypotonus improved after vibration stimulation; Romberg, Tandem: increase in score ≥was judged as effective; TUG: >10% decrease in elapsed time (10% faster) was judged as effective -, Impossible to do; BLS, Burke Lateropulsion Scale; Cont, contralesional; Lt, left; Rt, right; TUG, Timed-Up-and-Go. Among the three balance tests used in the present study, the pass rates of the tests (ie, number of patients who passed the test to the number assessed) were in descending order of Romberg test>tandem test>TUG. Results Patient profiles We studied a total of eight consecutive patients, including case 1 presented above, with a diagnosis of lateral medullary syndrome (Wallenberg syndrome), who were admitted to our facility for rehabilitation targeting gait disturbance or dysphagia. All patients had a definitive diagnosis based on clinical symptoms and neuroimaging findings. Patient profiles are shown in table 1. The patients comprised seven males and one female, aged 48 to 87 (mean 61.5±17.4) years. Duration from onset to admission ranged from 16 to 39 (mean 24±8) days, duration from onset to independent ambulation from 23 to 101 days (mean 52.5±27.3) days, and length of hospital stay from 24 to 136 (mean 72.9±40.6) days (table 1). Baseline function before stimulation At admission, all patients exhibited gait lateropulsion toward the lesional side. Seven patients were assessed as BLS score ≥2 (requires walking aid or needs assistance for standing and walking) while the remaining one patient was evaluated as BLS score 1 (self-awareness of mild lateropulsion but not requiring walking aid) (tables1 3). After admission, as a result of balance training conducted before the vibration stimulation study, four of eight patients (cases 2, 3, 4 and 5) achieved BBS scores ≥46,20 indicating independent walking on level surfaces (‘ambulator–independent level surface only’ classified by Functional Ambulation Category).21 For case 2, however, evaluation of independent walking was conducted 1 day before initiation of vibration stimulation. The temporal relationship between admission, vibration stimulation initiation, and independent walking achievement is shown in figure 2. These four cases were not included in the evaluation of the effect of hybrid training, but were included in the evaluation of immediate effect of vibration stimulation. Clinical courses from disease onset to independent walking and hospital discharge. An asterisk after the case number indicates that the case was studied prospectively for long-term effect. In cases 3, 4 and 5, which had achieved independent level walking early in rehabilitation, Romberg test had already reached score 4 at vibration stimulation initiation. In contrast, cases 1, 6, 7 and 8, which had not yet achieved independent walking at vibration stimulation initiation, showed marked balance dysfunction as indicated by both Romberg and tandem test scores <4 (except for case 8 with score 4 for Romberg test only). Time required to achieve independent walking was 49, 53, 29 and 8 days, respectively. In cases 1, 6 and 7, we conducted hybrid training for gait disturbance, and observed the course prospectively to evaluate the effectiveness of this approach in improving balance function. Case 8 was excluded from this analysis because of the short interval (8 days) between vibration stimulation initiation and independent walking. Baseline values before stimulation Among the 7 cases (cases 2–8) that were evaluated using three balance assessment tests, all but case 7 received stimulation on the lesional side first. The baseline values before lesional-side stimulation were as follows. For Romberg test, four patients were able to maintain a standing posture with eyes closed (score 4), while 3 patients scored <4. For tandem standing test, only one patient was able to maintain tandem stance for 15 s (score 4), and the remaining six patients were unable to assume the tandem stance (score 0). For TUG test, excluding two non-ambulant patients, the mean time for the remaining five patients was 10.7±1.9 s, which was shorter than the cut-off value (20 s) for locomotive instability defined by the Japanese Orthopaedic Association.22 At baseline before non-lesional-side stimulation, 6 of 7 patients were able to maintain a standing position with eyes closed (score 4) in Romberg test; one patient maintained the tandem stance for 15 s in tandem test (score 4), while six patients could not assume the tandem stance (score 0). The prestimulation baseline values of both Romberg and tandem tests were not significantly different between lesional-side and non-lesional-side stimulation when compared using Fisher’s exact test with significance level set at p<0.05 (Romberg test: p=0.5594, tandem test: p>0.9999). Therefore, comparison of the poststimulation effects of vibration between lesional-side and non-lesional-side stimulation was possible. Immediate effects of vibration stimulation Even though patient’s subjective observation was positive (see later), BLS-walking scores were not improved in 6/8 cases, except for two cases (patient 1 and 8, both improved from grade 2 (moderate) to 1 (mild lateropulsion)) (table 3). We examined whether vibration stimulation had an immediate effect in any of the three tests; Romberg, tandem standing test, and TUG test, regardless of whether the individual had already achieved independent level walking at the time of stimulation. No significant differences in scores were found for any of the three individual balance function tests before and after vibration (Wilcoxon matched-pairs signed rank test, figure 3). The same results were obtained for both lesional-side and non-lesional-side stimulations. Immediate effect of vibration stimulation. Numbers in the graph indicate the number of cases. Statistical data (Fisher’s exact test) are shown in the tables under the graphs. All comparisons between previbration and postvibration stimulation showed no significant difference with Wilcoxon (matched-pairs) signed-ranks test. L, lesional-side stimulation; NL, non-lesional-side stimulation; TUG, Timed-Up-and-Go. Vibration stimulation was considered to be effective if at least one test showed an improvement in score ≥1, or at least one test showing ≥10% improvement in walking speed after stimulation. For lesional-side stimulation, two of eight patients showed improvement in two tests and the other six patients in one test. Overall, all eight patients showed improvement in at least one of the three tests. In contrast, for non-lesional-side stimulation, only three of eight patients showed improvement in one test, while the other five patients showed no improvement. Fisher’s exact test showed a significant difference between lesional-side and non-lesional-side stimulations (p=0.0256). These findings indicated the immediate effect of vibration stimulation applied to the lower limb on the lesional side (p<0.05). In patients showing no improvement with non-lesional-side stimulation, this trial was performed 3–4 days after lesional-side stimulation. During this interval, some patients showed improvement in some of the tests, and the differences between lesional-side and non-lesional-side stimulations became undetectable. Specifically, case 2 improved in Romberg test conducted 3 days later; case 6 improved in Romberg test conducted 4 days later; and case 8 improved in tandem test conducted 4 days later. Of the test items used, Romberg test tended to show a ceiling effect, while many cases had low scores in the tandem test. Duration of vibration stimulation In case 1, stimulation for 5 min did not sufficiently improve gait disorder, while stimulation for an additional 5 min led to stable walking. In another case, prolonging the stimulation duration to longer than 5 min also achieved definitive effects. Therefore, vibration stimulation was typically administered for 10 to 15 min in this study. Duration of balance-improving effect In case 1, stimulation of the right lower limb improved walking speed and stability. The effect appeared to last throughout the day of stimulation, but disappeared the following day. In case 6, we measured how long the vibration-induced improvement in Romberg score (ability to maintain closed-eye standing posture with narrowed step width) lasted by repeating the test. We observed that the effect lasted until 40 min after vibration stimulation but disappeared thereafter. Patient’s subjective observation (introspection) regarding stimulation Patients remarked the following: ‘feeling of improved body balance’, ‘feeling of counteracting when the body tilted’, and ‘feeling of exerting force on the toes’, only when stimulation was administered to the lesional side. All patients were subjectively aware of the beneficial effects of vibration stimulation on impaired balance function. Order of applying vibration stimulation Based on patients’ subjective observation, we hypothesised that vibration stimulation may produce effects that are subjectively perceived by the body. We tested this hypothesis in case 7. Without providing prior information, we applied vibration stimulation to the non-lesional side and then conducted tests. Thereafter, we applied stimulation to the lesional side. The results are shown in table 3. Stimulation of the right lower limb (non-lesional side) resulted in no improvement in both the Romberg and tandem tests. After confirming that stimulation of the non-lesional side produced no effect, we stimulated the left lower limb (lesional side) several hours later on the same day. The patient reflected that he felt being able to exert force firmly on his left lower limb, and the Romberg test score improved from 0 to 4. Long-term effects of vibration stimulation We investigated whether hybrid training had long-term effect on improving balance disorders, based on the longitudinal changes in the scores of three tests. In this study, five of eight cases (cases 2, 3, 4, 5 and 8) were excluded because they were already walking independently or expected to become independent soon at vibration stimulation initiation. Case 1, the initial case of this study, was analysed for changes in walking-related factors over time. The remaining two cases (cases 6 and 7) were evaluated multiple times during hospitalisation (figure 2). Case 6 showed improvements in Romberg and tandem tests 10 days after starting hybrid training, with time in TUG test improving from 24.2 s to 23.5 and then 14.3 s. In case 7, the patient was able to undergo TUG test at 7 days after starting hybrid training, showing improvement in Romberg test at 14 days, and in tandem test at 21 days. TUG improved from 43.6 to 32.8 s. Case 1 did not undergo Romberg or tandem test but showed improvement in the time taken to walk 10 steps: from 22.7 s to 18.3, 16.7, and finally 10.8 s (online supplemental figure 1). Time to independent walking and discharge The average duration from onset to independent walking was 52.5±27.3 days. Vibration stimulation was initiated on average 14.6±12.1 days after admission. Of the eight patients, three were already walking independently at the initiation of stimulation, and one patient walked independent 1 day after starting stimulation. In four of eight patients, gait lateropulsion was resolved during vibration stimulation; instability was observed in Romberg and tandem tests, but improved with vibration stimulation (figure 3). In four of eight patients (cases 1, 6, 7 and 8), the mean duration from the start of vibration stimulation to independent walking was 26.4±21.9 days (range 8 to 53 days) (table 1). The duration from onset to discharge ranged from 43 to 153 days (average 101.3±43.3 days). Hospitalisation period tended to be longer in patients with swallowing disorder (cases 3, 4) and in a patient who wished to return to work (cases 5) (figure 2 and table 1). Discussion Epidemiology Vertebral artery dissection is a major cause of stroke in young and middle-aged individuals, accounting for 8%–10% of strokes in this group. In our study, four of eight subjects were diagnosed with vertebral artery dissection (see online supplemental text). Brainstem infarctions from vertebral artery dissection generally have favourable functional outcomes.23 In lateral medullary syndrome (Wallenberg syndrome) characterised by lesion location in the medulla, lesional-side trunk lateropulsion during walking is usually transient, typically resolving within weeks.6 9 24 25 However, gait lateropulsion may hinder social rehabilitation.26 In our cohort, four of eight patients continued to show impairments in Romberg and tandem tests, despite achieving independent level walking, suggesting persistent balance disorders. Duration until discharge was influenced by gait stability and the need to rehabilitate for dysphagia and balance impairment not evident during level ambulation. Baseline comparison for lesional-side versus non-lesional-side stimulation To confirm the effectiveness of vibration stimulation on the lesional side, baseline balance function must be equivalent between lesional and non-lesional stimulations. In three cases, baseline test scores were slightly better for non-lesional stimulation, probably due to continuous rehabilitation during the 3–4-day interval between the two stimulations, possibly contributing to improved balance ability. However, statistical analysis revealed no significant baseline differences between lesional and non-lesional-side stimulations. Immediate effects Balance performance (Romberg or tandem test) or gait improved in all eight cases following lesional-side vibration stimulation, compared with only three cases after non-lesional-side stimulation, with a significant difference. Thus, vibration stimulation of lower limb muscles on the lesional side for ≥10 min effectively achieved immediate improvement of balance impairment in Wallenberg syndrome. Long-term effects Three patients demonstrated balance improvements during hybrid training. As lateropulsion was transient and resolved gradually, and some tests improved within a few days, also vibration effects were self-limited for several ten minutes, superiority of hybrid training over conventional balance training could not be established. Nevertheless, hybrid training may be recommended in cases of sustained lateropulsion. Causes of lateropulsion and treatment Lateropulsion has been reported to involve the vestibulospinal tract, vestibular nucleus, posterior spinocerebellar tract, cerebellum and medial lemniscus.139 Treatments for gait lateropulsion targeting vestibulospinal tract involvement include subjective visual vertical training26 and direct current stimulation.27 For spinocerebellar tract involvement, compensatory use of somatosensory input,11 12 28 visual feedback29 and therapies using weights or elastic bandages10 13 14 have been reported. However, evidence remains limited to case reports, with no large series available. Mechanism of vibration therapy Perception of limb motion involves kinesthesia—conscious sense of movement—and unconscious proprioceptive input from peripheral receptors.30 Both arise from input through peripheral sensory receptors. It is thought that kinesthesia is transmitted via dorsal columns-medial lemniscus pathway, while proprioceptive input is transmitted through the spinocerebellar tract. It is now known proprioceptive information travelling dorsal spinocerebellar tract relays at nucleus Z in medulla, through thalamus and, as kinesthesia, up to sensorimotor cortex.31 Vibration stimulation excites Ia nerve fibres by activating muscle spindles.32 This triggers the tonic vibration reflex via anterior horn cells of homonymous muscles in the spinal cord, and inhibits antagonist muscles through reciprocal Ia inhibition. Devices applying vibration stimulation to paretic hands of hemiplegic patients to facilitate motor functions have been reported.33 Functional MRI has shown that 100 Hz vibration stimulation increases blood flow across wide CNS regions, including the cerebellar vermis region during motor tasks.34 In seven of eight patients, acute symptoms such as dizziness, vomiting, and nystagmus had resolved by the time of rehabilitation initiation. All cases studied exhibited sensory disturbance patterns typical of lateral medullary syndrome, without unilateral deficit in tactile recognition or kinesthesia including sense of joint position. Therefore, unilateral lateropulsion was primarily due to lesions in the spinocerebellar tract rather than the vestibulospinal tract or medial lemniscus. Vibration stimulation applied to the lesional-side lower limb should reach the ipsilateral cerebellar vermis via the ipsilateral spinocerebellar tract. If this pathway is blocked, from where do the effects of vibration stimulation originate? For the posterior pathway (the dorsal spinocerebellar tract), neurons from muscle spindles in the lower limb project to Clarke’s column in the spinal cord, and the axons travel ipsilaterally via the inferior cerebellar peduncle to terminate in the ipsilateral cerebellar vermis. For the anterior spinocerebellar tract, information from the lower limbs enters the spinal cord and crosses to the contralateral side, ascends the contralateral spinal cord, and descends from the superior cerebellar peduncle to terminate bilaterally in the cerebellar vermis.243537 Information from spinal central pattern generators related to gait also transmits through the anterior tract.38 Lateropulsion in our patients likely resulted from loss of ipsilateral proprioceptive input due to lesional-side posterior tract interruption. Recovery from vestibulospinal lesions relies on central compensation for the imbalance between the lesional and non-lesional sides.24 37 If the anterior spinocerebellar tract remained intact, vibration stimulation could have contributed to central compensation. A study suggests that human posterior spinocerebellar pathways differ from animal models.39 The effects of vibration stimulation on lateropulsion mediated by pathways other than the posterior spinocerebellar tract, especially when the tract is disrupted, remain unclear and require further investigations. Effects of lesional versus non-lesional stimulation This study found significantly greater effects when vibration was applied to the ipsi-lesional side compared with the contralesional side. Stimulation ≥10 min is required; 5 min was insufficient. Effects lasted 30 to 40 min, and disappeared the following day. Effects of continuous vibration on the spinal cord are known to last approximately 20 min, suggesting possible physiological actions exerted by vibration stimulation.32 Patient’s subjective observations Patients frequently reported subjective sensations of improved force exertion or posture, suggesting effects on muscle strength or muscle tone. Although our single-case observations cannot establish benefits of vibration stimulation in permanent balance improvement, the subjective sensory changes support benefits in motor learning. Further research is needed to determine how such transient phenomena contribute to rehabilitation outcomes. Limitations This study has several limitations. First, in the present study, the effect of balance function improvement following vibratory stimulation to whole leg muscles was widespread, sustained, and effective on the lesioned side—features suggesting involvement of the ascending tracts on the lesion side. However, local effects on the stimulated muscle due to tonic vibration reflex could not be ruled out. Second, the administration sites of vibration stimulation were not specifically defined, which may have increased the variability of the results. Third, the patients were not examined for vestibular function because vestibular symptoms were not observed at baseline. However, seven of eight patients had early acute symptoms including dizziness, vomiting and nystagmus, although they were resolved by the time rehabilitation started. Since these were also common symptoms observed in vestibular involvement,4 40 vestibular involvement at the early stage cannot be ruled out completely in our patients. Fourth, objective comparison of degree of improvement between patients is difficult (eg, a patient who progressed from unable to assume tandem stance to maintaining the posture for 10 s vs another patient who doubled Romberg test performance). Fifth, stimulation order of lesional-side and non-lesional-side stimulations was not randomised. Seven of eight patients received lesional-side stimulation first, raising the possibility of examiner bias despite efforts to minimise it. Finally, we conducted a single-case observation study, which cannot establish long-term effects because of rapid improvements in balance tests. We assumed the high-volume balance training as a cause of improvement, but it is necessary to consider the patients were in the process of natural recovery. Larger randomised trials with patients matched for severity of balance impairment are required to determine whether vibration stimulation yields sustained benefits. Supplementary material