Bilateral Transcutaneous Auricular Nerve Stimulation Promotes Motor Function Recovery After Stroke: A Randomized Controlled Trial

Clinical Question (PICO)

In adults aged 18–65 years at least 2 weeks after ischemic or hemorrhagic stroke, with stable vital signs and unilateral motor dysfunction affecting the upper and lower limbs (P), does bilateral transcutaneous auricular nerve stimulation (BtaNS) added to standardized rehabilitation (I), compared with left-sided stimulation (LtaNS) or sham stimulation (C), improve upper- and lower-limb motor function, balance, and activities of daily living through 8 weeks after treatment (O)? The exploratory subgroup questions examined age, time since stroke onset, and stroke etiology.

Bottom Line

In this single-center randomized trial, bilateral stimulation plus routine rehabilitation produced the largest gains across motor, balance, and daily-living measures: at 4 weeks, FMA-UE was 45.7 versus 41.0 with LtaNS and 33.4 with sham, while WMFT score was 56.1 versus 52.1 and 43.8, respectively (omnibus P<0.001 for each table comparison). At week 12, BtaNS also had higher mean BBS and Modified Barthel Index scores than LtaNS or sham. The signal is promising but not practice-changing: only 102 of 111 randomized participants contributed to the per-protocol behavioral analysis, and the 0-mA control could reveal allocation because active stimulation produced tingling.

Design

Trial type: Prospective, double-blind, three-arm, single-center randomized controlled trial.
N: 111 randomized; 102 completed behavioral follow-up and were analyzed per protocol.
- BtaNS: 37 randomized; 34 analyzed.
- LtaNS: 37 randomized; 34 analyzed.
- Control/sham: 37 randomized; 34 analyzed.
Randomization: 1:1:1 stratified block randomization by stroke type, with fixed blocks of six and sequentially numbered opaque sealed envelopes.
Setting: Department of Rehabilitation Medicine, First Affiliated Hospital of Nanchang University, China; inpatient rehabilitation.
Enrollment: January–December 2025.
Mean follow-up: Assessments at baseline, 2 weeks, 4 weeks, and week 12 (8 weeks after treatment ended).
Analysis: Repeated-measures ANOVA with Bonferroni correction when assumptions were met; GEE or nonparametric methods otherwise. The final behavioral analysis was explicitly per protocol, excluding withdrawals and losses to follow-up; fNIRS analysis was descriptive.
Primary outcome: FMA-UE (0–66), FMA-LE (0–34), and WMFT score/time, assessed at 2 and 4 weeks and follow-up.

Population

Inclusion Criteria

Age 18–65 years.
Ischemic or hemorrhagic stroke diagnosed according to the 2021 Chinese stroke guidelines.
Recovery phase at least 2 weeks after onset, with stable vital signs and no ongoing progression.
Unilateral motor dysfunction affecting both upper and lower limbs.
Able to participate in standardized rehabilitation and provide informed consent.

Exclusion Criteria

Motor dysfunction from a non-stroke cause.
Severe cognitive impairment or psychiatric illness preventing treatment compliance.
Unstable vital signs or recurrent stroke at screening.
Implanted pacemaker or other contraindicating medical device.
Uncontrolled epilepsy, vasovagal syncope, or another contraindication to nerve stimulation.

Baseline Characteristics (overall or representative arm)

Mean age was 54.3, 53.6, and 52.1 years in the control, LtaNS, and BtaNS groups, respectively; 102 participants had complete behavioral data.
Hemorrhagic stroke comprised 64.7%, 55.9%, and 58.8% of the three groups; ischemic stroke comprised 35.3%, 44.1%, and 41.2%.
Mean baseline FMA-UE was 29.2, 29.7, and 29.9; mean FMA-LE was 16.8, 16.9, and 16.6; baseline WMFT score was 38.5, 39.0, and 38.5.
Baseline BBS and Modified Barthel Index were approximately 31 and 70, with no significant between-group differences.

Interventions

BtaNS: Bilateral auricular stimulation through electrodes at the cymba and cavum conchae; 20 Hz, 300 µs square-wave pulses, individually tolerated intensity 0.5–1.0 mA, 30 minutes twice daily, 6 days/week for 4 weeks, immediately followed by 30 minutes of routine rehabilitation.
LtaNS: The same schedule and device, with active stimulation at the left ear and sham stimulation at the right ear.
Control/sham: The same headset, electrode placement, duration, and rehabilitation schedule, but 0 mA on both sides.
Routine rehabilitation: Task-oriented training, postural control, balance, gait, and sensory-integration therapy, delivered twice daily by experienced physiotherapists; rehabilitation dose was intended to be equal across groups.

Outcomes

Primary Outcome (FMA-UE, FMA-LE, and WMFT):

Baseline values were similar. At 2 weeks, BtaNS/LtaNS/control means were FMA-UE 39.4/34.1/31.0, FMA-LE 24.1/21.1/18.7, WMFT score 48.1/45.8/41.4, and WMFT time 146.2/187.9/260.5 seconds. The reported omnibus P values were <0.001 for these comparisons; the article did not provide numeric pairwise 95% CIs in the text.
At 4 weeks, BtaNS/LtaNS/control means were FMA-UE 45.7/41.0/33.4, FMA-LE 27.1/23.7/20.1, WMFT score 56.1/52.1/43.8, and WMFT time 70.3/138.5/249.4 seconds; reported omnibus P values were <0.001 for each comparison.
The BtaNS advantage over LtaNS was reported as significant at the measured time points for most outcomes; the WMFT-score difference between BtaNS and LtaNS was not significant at 2 weeks but was favored BtaNS at 4 weeks. No numeric pairwise effect estimates or 95% CIs were reported in the article text.
Subgroups (exploratory/post-hoc): WMFT-time benefit was greater with disease duration ≤1 month and in hemorrhagic stroke, with interaction P=0.006 and P=0.018, respectively. Age did not materially alter the treatment pattern. For WMFT score, interaction P values were 0.890 for age, 0.511 for disease duration, and 0.165 for etiology, supporting broadly consistent score gains. Figure 7 displays treatment differences with 95% CIs, but the paper does not tabulate their numeric values.

Secondary Outcomes:

At 4 weeks, BtaNS/LtaNS/control mean BBS was 46.5/42.4/36.8 and mean Modified Barthel Index was 83.6/77.2/74.4; the reported omnibus P values were <0.001 for both.
At week 12, mean BBS was 49.6/45.1/38.6 and mean Modified Barthel Index was 85.8/79.6/75.5 for BtaNS/LtaNS/control, with reported omnibus P values <0.001.
The week-12 low-fall-risk proportion was 91.18% with BtaNS, 70.59% with LtaNS, and 44.12% with control. Full independence on the Modified Barthel Index was reported in 8.82% of BtaNS participants and none in the other groups.
Exploratory fNIRS in only 6 BtaNS, 7 LtaNS, and 6 control participants showed more widespread bilateral HbO changes after BtaNS; no formal statistical inference was performed.

Adverse Events / Safety:

No serious adverse event attributable to stimulation was identified, and heart rate and blood pressure remained within normal clinical ranges.
Ear pain occurred in 2 control, 5 LtaNS, and 6 BtaNS participants; ear pruritus in 3, 4, and 5; and ear redness in 0, 1, and 2, respectively. Group differences were not significant.
One control participant had deep-vein thrombosis and one LtaNS participant had a seizure; the authors considered these unrelated to stimulation. No event required additional treatment or caused treatment discontinuation because of stimulation.

Figures

*Figure 1.* Clinical trial procedures and assessment time points. T0: Baseline level. T1: During treatment at 2 weeks (Week 2). T2: End of treatment at 4 weeks (Week 4). T3: Follow-up at 8 weeks after the end of treatment (Week 12).

Bilateral transcutaneous auricular nerve stimulation device and sites — *Figure 2.* The schematic diagram of bilateral transcutaneous auricular nerve stimulation implementation. (a) BtaNS device; (b) stimulation sites and electrodes. BtaNS: bilateral transcutaneous auricular nerve stimulation.

CONSORT flowchart of the randomized trial — *Figure 3.* The CONSORT flowchart of the study. LtaNS: left-sided transcutaneous auricular nerve stimulation; BtaNS: bilateral transcutaneous auricular nerve stimulation.

Cardiovascular measurements during auricular stimulation — *Figure 4.* The effects of LtaNS and BtaNS on cardiovascular measurements in stroke patients. Measurements were taken daily over the 24 days of the experimental therapy. (a) HR; (b) SBP and (c) DBP. Data are expressed as mean ± SD and were analyzed by repeated-measures analysis of variance. DBP: Diastolic blood pressure; HR: heart rate; post: post-treatment; pre: pre-treatment; SBP: systolic blood pressure; LtaNS: left-sided transcutaneous auricular nerve stimulation; BtaNS: bilateral transcutaneous auricular nerve stimulation.

Neurological function scores after treatment — *Figure 5.* Changes in neurological function scores of stroke patients. (A–F) FMA-UE score, FMA-LE score, WMFT score, WMFT time, Berg Balance Scale (BBS), and Modified Barthel Index (MBI) at baseline (Pre), 2 weeks, 4 weeks, and 8-week follow-up. Data are expressed as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 versus LtaNS group or #P < 0.05, ##P < 0.01, ###P < 0.001 versus LtaNS group (repeated-measures analysis of variance). FMA-UE: Fugl–Meyer Assessment–upper extremity; FMA-LE: Fugl–Meyer Assessment–lower extremity; WMFT: Wolf Motor Function Test; BBS: Berg Balance Scale; MBI: Modified Barthel Index; LtaNS: left-sided transcutaneous auricular nerve stimulation; BtaNS: bilateral transcutaneous auricular nerve stimulation.

Balance and Modified Barthel Index categories at week 12 — *Figure 6.* Distribution of Berg Balance Scale and Modified Barthel Index categories at week 12. (a) Distribution of Berg Balance Scale (BBS) categories at week 12. Blue indicates medium fall risk and green indicates low fall risk; (b) Distribution of Modified Barthel Index (MBI) categories at week 12. Red indicates moderate dependence, blue indicates slight dependence, and green indicates independence. The proportions of patients (%) are shown for the control group, LtaNS group and BtaNS group. Low fall risk (BBS ≥ 41), median fall risk (21≤BBS ≤40), moderate dependence (61≤MBI≤90), slight dependence (91≤MBI≤99), dependent (MBI=100). LtaNS: left-sided transcutaneous auricular nerve stimulation; BtaNS: bilateral transcutaneous auricular nerve stimulation.

Subgroup analyses of WMFT outcomes — *Figure 7.* The subgroup analysis result of WMFT score and WMFT time. (a) Subgroup analyses of the WMFT time; (b) Subgroup analyses of the WMFT score. Subgroup analyses were performed according to age, disease duration, and stroke etiology. Data are presented as treatment differences with 95% confidence intervals. P values for interaction are shown to assess heterogeneity of treatment effects across subgroups. LtaNS: left-sided transcutaneous auricular nerve stimulation; BtaNS: bilateral transcutaneous auricular nerve stimulation; WMFT: Wolf Motor Function Test.

Cortical hemodynamic change maps — *Figure 8.* Brain hemodynamic change maps comparing the LtaNS and BtaNS groups with the control group before and after treatment. Cortical hemodynamic maps are shown for the control, LtaNS, and BtaNS groups at baseline and post-intervention. Color-coded surface maps represent the magnitude of HbO signal changes, with warmer colors indicating higher values and cooler colors indicating lower values, as shown in the color bar. All maps are displayed on a standardized cortical surface to facilitate visual comparison across groups and time points. LtaNS: left-sided transcutaneous auricular nerve stimulation; BtaNS: bilateral transcutaneous auricular nerve stimulation.

Proposed mechanisms of bilateral auricular stimulation — *Figure 9.* Proposed mechanisms underlying the effects of bilateral taNS on post-stroke motor recovery. M1: Primary motor cortex; SMA: Supplementary motor area; PMC: Premotor cortex; FMA-UE: Fugl–Meyer Assessment–upper extremity; FMA-LE: Fugl–Meyer Assessment–lower extremity; WMFT: Wolf Motor Function Test; BBS: Berg Balance Scale; MBI: Modified Barthel Index; fNIRS: functional near-infrared spectroscopy; taNS: transcutaneous auricular stimulation.

Criticisms

The randomized cohort was small, single-center, and restricted to adults 18–65 years who could complete inpatient rehabilitation; external validity to older, severely impaired, medically unstable, or community-dwelling stroke survivors is uncertain.
The primary behavioral analysis excluded 9 randomized participants and used a per-protocol rather than intention-to-treat approach, creating potential attrition and adherence bias.
The control group received 0 mA and therefore lacked the tingling sensation of active treatment. Participants could plausibly infer allocation despite the claimed double-blind design; an active sensory sham is needed.
Ischemic and hemorrhagic stroke were pooled. The reported etiology and disease-duration subgroup findings were exploratory, with limited detail and no numeric effect estimates or confidence intervals in the text.
Repeated testing across several scales and time points increases multiplicity concerns. Statistical significance is not the same as a patient-important change, and the paper does not anchor the between-group differences to a prespecified minimal clinically important difference.
fNIRS findings came from 19 participants selected partly by feasibility and were explicitly descriptive; they cannot establish the proposed vagal or cortical mechanism.
Follow-up ended 8 weeks after the 4-week intervention, which is inadequate to establish durable motor-network change or sustained independence.

Funding

The study was supported by the Clinical Research Incubation Program at the First Affiliated Hospital of Nanchang University (YFYLCYJPY202450), the Natural Science Foundation of Jiangxi Province (20252BAC240542), and the Research Program for Postgraduate Students of the Jiangxi Association for Science Education (025KXJYS228). The article reports registration in the Chinese Clinical Trial Registry as ChiCTR2400093825. No commercial device funding is reported in the funding statement.

The paper

Authors. Wang J, Liu Q, Zhou Y, Peng C, Dong X.
Title. Bilateral transcutaneous auricular nerve stimulation promotes motor function recovery after stroke: A randomized controlled trial.
Journal. Therapeutic Advances in Neurological Disorders.
Year. 2026.
DOI. 10.1177/17562864261476124
PMCID. PMC13443378

Deep Dive — click to expand

What this is

This is a small, single-center rehabilitation trial asking whether bilateral non-invasive auricular stimulation adds anything to standardized post-stroke therapy beyond unilateral stimulation or sham. The headline is a coherent dose-and-distribution signal: after four weeks, BtaNS participants had higher FMA, WMFT, balance, and Barthel scores, and the apparent benefit persisted to week 12. The catch is equally important: the clinical results are completer-only, the sham was electrically inert rather than sensation-matched, and the mechanistic fNIRS sample was tiny.

1. Shadow Audit

The paper foregrounds “double-blind” randomization, but the control condition was 0 mA while active stimulation was 0.5–1.0 mA and intentionally perceptible. That is a practical unblinding channel. The nine participants who withdrew or were lost to follow-up were removed from the final behavioral analysis, leaving 34 per group from 37 randomized; the analysis therefore cannot preserve the protection that randomization provides against differential dropout.

There is also a scale-level framing issue. At week 12, the BtaNS group’s mean BBS was 49.6 versus 45.1 for LtaNS and 38.6 for control, and the mean Barthel Index was 85.8 versus 79.6 and 75.5. Those are encouraging distributions, but the trial does not show whether participants crossed a validated patient-important threshold because of BtaNS, nor whether gains persisted beyond eight weeks after stimulation stopped. “Bilateral cortical engagement” is a plausible hypothesis, not a demonstrated mediator: the fNIRS subset was only 19 people and had no formal inferential analysis.

2. Inversion Engine

The conclusion would invert if the apparent BtaNS-specific effect were mainly expectancy, sensation, selective retention, or unusually responsive patients rather than bilateral stimulation. The observed week-4 BtaNS-versus-LtaNS separations were 4.7 FMA-UE points, 3.4 FMA-LE points, 4.0 WMFT-score points, and 68.2 seconds on WMFT time; after an active-sensation sham and intention-to-treat analysis, those differences would need to shrink toward zero or lose consistency across outcomes for the bilateral claim to fail. The paper does not report a validated minimal clinically important difference, so a more precise bedside inversion threshold cannot be responsibly supplied from this study.

The decisive experiment is a multicenter factorial-quality trial with identical sensation in BtaNS, LtaNS, and control, blinded allocation checks, intention-to-treat analysis, prespecified stratification by stroke type and recovery stage, and a 3–6-month functional endpoint. If the week-12 BBS, Barthel, and WMFT advantages remain with confidence intervals excluding no difference, the inversion becomes unlikely; if only the unblinded per-protocol scales remain positive, the current interpretation collapses.

3. Second-Order Catalyst

If this result is real, the first mover is an inpatient stroke-rehabilitation service—not the hyperacute stroke bay. The intervention is inexpensive and operationally simple: two 30-minute stimulation sessions per day, six days per week, for four weeks, each immediately paired with routine motor therapy. A service could begin with a research protocol tomorrow: document baseline FMA/WMFT/BBS/Barthel scores, use an active sensory sham, record adherence and allocation guesses, and stratify the audit by hemorrhagic versus ischemic stroke and by time since onset.

The subgroup signal suggests the earliest practical target would be patients within one month of stroke, but it is exploratory and should guide trial enrichment rather than justify routine selection. The protocol should not displace task-specific therapy; the paper’s intervention was an adjunct to an equal rehabilitation dose.

4. Asymmetric Leverage

The upside is asymmetric if a low-cost, non-invasive add-on can reliably improve the function of a large rehabilitation denominator. A four-week course delivers 48 sessions, and a modest improvement in transfer, gait, hand use, or self-care across many patients could have more system value than a dramatic result in a tiny specialist subgroup. The safety table is reassuring enough to make that hypothesis testable: no serious stimulation-attributable events and only mild ear symptoms were observed.

The asymmetry cuts the other way on evidence quality. A 102-person completer analysis can generate a large-looking signal across six functional measures without establishing durability, MCID attainment, or real-world benefit. The leverage is therefore in a definitive pragmatic trial with an honest sham, not in immediately purchasing devices for routine deployment.

5. Paradigm Destroyer

This paper challenges two reflexes: that neuromodulation must be implanted to matter, and that one-sided stimulation is automatically the safest or most rational auricular default. It does not, however, kill the more important reflex of demanding a credible sham and an intention-to-treat analysis.

The protocol update for tomorrow morning should be: “Treat auricular stimulation as an adjunctive rehabilitation research option, never as a replacement for task-specific therapy; if offered, use sensation-matched sham controls and record function at baseline, four weeks, and at least three months.”

MVP — Minimum Viable Proof

The minimum bedside-changing statement would be: in a multicenter intention-to-treat trial of a similar post-stroke rehabilitation population, BtaNS preserves a clinically meaningful, prespecified advantage over a sensation-matched sham on at least one validated motor or daily-function endpoint at week 12, with a 95% confidence interval excluding no difference, without an excess of serious adverse events or differential dropout. This paper supplies a candidate effect signal—roughly 4.7 FMA-UE points over LtaNS at week 4—but not the sham, durability, or analysis quality needed to call it proof.

Best Combination

Combine this signal with the prior implanted-VNS rehabilitation literature and existing unilateral taNS studies as a hypothesis about pairing neuromodulation with active motor practice, not as evidence that bilateral ear stimulation is already equivalent to an established therapy. The clean synthesis is: non-invasive stimulation may be a scalable adjunct; bilateral dosing may be worth testing; the next study must separate stimulation biology from sensation, attention, therapist enthusiasm, spontaneous recovery, and differential completion.

Overvalue Warning

First, do not turn the fNIRS maps into proof that the vagus nerve caused the recovery. The authors themselves note that auricular innervation is complex and not exclusively vagal, and the fNIRS sample was only 19 participants without formal statistical inference.

Second, do not quote the 91.18% low-fall-risk proportion or the BtaNS mean Barthel score as if they were independently adjudicated, durable patient outcomes. They came from 34 completers per group, after a potentially unmasking sham, and the trial did not establish whether the scale changes were clinically meaningful or persisted long term.

Educational use notice. The Time Is Brain Review is dedicated to medical education. Figures from published articles are embedded for teaching purposes, with attribution to the original publication. We do not claim copyright over any embedded figure; all rights remain with the original publisher and authors under the article's published license. If you are a rights holder and would like a figure removed, please contact the editorial team. See our editorial notice for the full policy.