Participants

This is a single-centre, randomised, double-blind study. The study includes two substudies: an acute period (AP) study and a preventive period (PP) study, each with a taVNS group and a sham taVNS group. The AP study will assess the efficacy of taVNS in treating primary headaches during the acute stage, while the PP study will evaluate the preventive efficacy of taVNS on primary headaches. Participants eligible for this trial must meet several criteria, including a verified diagnosis of migraine, CH or TTH by a certified clinician. The diagnosis must adhere to the standards outlined in the 2022 editions of the Chinese Guidelines for the Diagnosis and Treatment of Migraine and Cluster Headache. Participants will undergo cranial MRI or CT scans to exclude headaches caused by organic brain lesions. Given the limited research on primary headache among the young population, as identified in a recent large meta-analysis,10 this study appropriately broadens the age range to include patients aged 7–20 years. Therefore, this study appropriately expanded the age range to include patients aged 7–20 years. Additionally, candidates should have endured headaches for 3–15 days per month in the preceding year, adhered to a stable regimen of medication in terms of dosage and frequency and will not take new medications throughout the trial. Comprehensive information about the study will be provided, and informed consent (see online supplemental appendix 1) must be obtained from both the patients and their legal guardians.

The study will exclude individuals with conditions that may interfere with outcomes or pose risks to participants. Exclusion criteria include secondary headaches and severe medical conditions such as aneurysms, intracranial haemorrhage, brain tumours or significant head injuries. Individuals with a history of drug abuse, addiction, syncope, seizures, foramen ovale occlusion or cervical vagotomy are also ineligible. Patients using opioids more than two times per month, analgesics or non-steroidal anti-inflammatory drugs (NSAIDs) more than 15 days per month or specific medications such as tamoxifen, ergots or combined analgesics more than 10 days per month will be excluded. Additionally, the study will not include children under 6 years of age, pregnant women or individuals with significant cardiovascular conditions, mental or cognitive disorders or contraindications to taVNS (eg, ear inflammation, brain implants or a history of epilepsy).

This study has received approval from the Medical Ethics Committee affiliated with the Nanfang Hospital of Southern Medical University for Clinical Studies (NFEC-2024–057). It has been registered on ClinicalTrials (NCT06277063) on 10 March2024.

Patient and public involvement

The parents of two children with migraine were involved in the protocol design. They provided practical advice on how to streamline the process of bringing the device home and teaching the participants and their parents to use it. Based on their feedback, several changes were made to the original study procedure, including presetting the parameters, simplifying the device return process and improving the method for recording VAS scores.

Sample size

The sample size was determined using n=2*[(Z_α+Z_β)^σ/d]² and statistical power analysis software G*Power 3.1. Since no prior studies have evaluated the effects of taVNS on paediatric populations with primary headaches, we estimated the sample size for the AP study based on data from a non-invasive VNS acute treatment study on adult primary headache by Martelletti et al,46 and for the PP study, based on the preventive treatment studies of taVNS by Zhang et al29 and another non-invasive VNS study by Diener et al.47 The effect size is estimated to range from 0.5 to 0.65 (Cohen’s d). For the AP study, we used a medium effect size (Cohen’s d=0.5), calculated from the difference in pain relief between the VNS and sham groups in Martelletti et al’s study and applied the parameters of α=0.05 and β=0.80. Based on these calculations, the estimated sample size for each group in the AP study was 64. For the PP study, the sample size was conservatively estimated based on a medium effect size (Cohen’s d=0.5) and a 20% dropout rate due to the 8-week treatment and follow-up period. A total of 160 participants will be enrolled.

Randomisation method and blinding

Patient candidates may opt to participate in either the AP or PP study after receiving a detailed briefing on the research protocol. On enrolment, participants will randomly select a sealed envelope containing a unique digital identifier associated with either the AP (1–128) or PP (1–160) study. In the AP study, participants drawing identifiers AP1–AP64 will be assigned to the taVNS group, while those with identifiers AP65–AP128 will be placed in the sham group. Similarly, in the PP study, participants drawing identifiers PP1–PP80 will be allocated to the taVNS group, and those with identifiers PP81–PP160 will be assigned to the sham group. To maintain allocation concealment, the contents of the envelopes will only be revealed after the follow-up assessments are completed. The entire randomisation process will be overseen by a research assistant who is not involved in the intervention or data analysis phases.

To maintain blinding, a certified doctor will screen eligible patients, remaining unaware of their group assignments. Second, enrolled participants will be randomly assigned to groups by a designated researcher. This researcher will receive an instruction sheet containing the digital identifier and corresponding stimulation site information without knowledge of the active or sham assignment. Using identical equipment for both groups (eg, EMG, ECG and stimulation devices), the researcher will collect biological signals and adjust stimulation intensity. Importantly, the researcher’s role will be limited to assisting and instructing participants and their parents on how to place the electrodes on the ear, without knowledge of which electrode sites correspond to the active or sham group. Both groups will experience a similar sensation of electrical stimulation at the electrode site, such as maximum tolerable stimulation without painful feeling, ensuring participant blinding. Third, all collected data will be anonymised before being provided to the data analysis team. Unblinding will occur only after the completion of all data analyses for participants, and this process will be conducted by a separate set of researchers to preserve blinding throughout the trial.

Procedures

AP study

As shown in figure 1A, the whole study process will mainly include three phases after enrolment: pretest (on Day 1), intervention (on Day X) and post-test (on Day 30±7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Study flow chart. (A) The timeline and procedures of the AP study. (B) The timeline and procedures of the PP study. (C) The 5–10–5 measurement during pretest and post-test for both AP and PP studies. AP, acute period; EMG, electromyography; PP, preventive period; taVNS, transcutaneous auricular vagus nerve stimulation.

Pretest period

On the first day after enrolment, participants will provide basic demographic information, including gender, age, medical history, family history of genetic conditions, current medical history, medication use and allergy history, and they will be assigned their randomisation number. Baseline data will then be collected, including VAS scores, EMG and ECG signals. EMG/ECG signals will be collected in three stages: a 5-min resting state, a 10-min taVNS/sham stimulation stage, and another 5-min resting state (5–10–5 measurement) (as shown in figure 1C). Specifically, the participant will be positioned in front of a computer, with both upper limbs comfortably placed on the table. They will then proceed to wear the ear electrodes, neck EMG sensors and the ECG sensor. Ear electrodes will be placed according to their grouping. The experimenter will disinfect the cymba concha area or the sham stimulation area of the participant’s left ear (as shown in figure 2A) using an alcohol wipe. Subsequently, the electrodes will be attached to the left ear. The amplitude of the taVNS device will be adjusted to ensure it remains below the participant’s pain threshold. The researcher will attach two sets of wireless EMG electrodes to the trapezius muscles on both sides of the neck to record surface EMG activity. A set of wireless ECG sensors will be placed on the chest to record ECG activity prior to the treatment sessions (as shown in figure 2B,C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

(A) The upper ear illustrates the electrode placement sites for participants in the taVNS group, with electrodes positioned on the cymba concha and earlobe. The lower ear depicts the electrode placement for the sham group, where electrodes are similarly positioned but placed near the auricle and earlobe. (B) Two pairs of wireless EMG electrodes will be attached to the trapezius muscles while the participant is in a supine position. (C) The positions of a pair of ECG electrodes are at RA and LL. EMG, electromyography; LL, left leg; RA, right arm; taVNS, transcutaneous auricular vagus nerve stimulation.

Following the baseline measurement, participants and their legal guardians will receive instructions on the proper use of the device. Once they are familiar with its operation, they will take the device home and use it at the onset of a headache attack. The stimulation parameters (except for the amplitude) of the taVNS device will be preset before it is taken home.

Period of intervention

When participants experience an acute headache, they will use the device for headache intervention according to the research staff training procedure. Simultaneously, the severity of the headache attacks will be recorded in a headache diary card by the VAS score. The heart rate and timing of the attacks will also be recorded. The treatment process involves administering stimulation within 20 min of onset, with a duration of half an hour, followed by immediate post-treatment assessments. Subsequent assessments will be conducted at 2 hours, 8–12 hours, 24 hours and 36–48 hours after the completion of the treatment.

Post-test period

The device will be returned to the hospital within 48 hours after completion of the intervention, and another 5–10–5 EMG/ECG measurement will be performed.

Outcome measures

The primary outcome measures and secondary outcome measures of two studies are shown in table 1.

For the AP study, the primary outcome measure is reduction in pain intensity, as assessed by VAS, 2 hours after the taVNS or sham treatment at home. The VAS is a widely used assessment tool that has been applied to various pain evaluations. In addition to the primary outcome, the AP study will include several secondary measures, including the change in VAS scores at other time points and the EMG and ECG features. Specifically, the averaged signal and the power density of EMG signals and the time domain and frequency domain of HRV features will be extracted.

For the PP study, the primary outcome measure is the reduction in the headache attack days during the intervention period. The secondary measures include mean headache attack days during the open-label period, the days of acute headache medication, mean headache intensity, mean duration of headache and the features derived from EMG and ECG.

Equipment and parameters

The taVNS device (BC102), EMG (BC107) and ECG (BC116) used in this experiment will be provided by Shenzhen BrainClos Technology Co. The stimulation parameters of the taVNS device are as follows: a frequency of 1 Hz, a pulse width of 300 μs and a stimulation intensity adjustable between 0 and 7.2 mA. The device delivers a constant current output, with the intensity finely adjustable across 60 levels, each representing a 0.12-mA increment. The output maintains a precision margin within 20%. The stimulation intensity will be customised based on the participant’s tolerance and comfort level. Both EMG and ECG have a sampling rate of 1000 Hz, and the signals are obtained through a 24-bit AD converter. The two devices are designed to use Bluetooth communication with very low noise to acquire uV-level electrophysiological signals. Bluetooth minimises signal artefacts from patient movements and avoids industrial frequency interference at 50 Hz.

EMG and ECG data processing

EMG signals filtered with a low-pass filter of 250 Hz. The filtered EMG signals were rectified and averaged in a moving window with a length of 200 samples. The power spectral density of 0–160 Hz of the signal will be extracted. Analyses were performed on representative data segments of 20-s length and the power density of representative frequency band of the EMG signals from resting stage, stimulation stage and resting stage during the EMG examinations of preintervention and postintervention, respectively.

ECG signals will be filtered with a low-pass filter of 20 Hz before identifying the R–R interval and calculating HR and HRV. HRV will be represented in both the time domain and frequency domain. The chosen index of HRV in the time domain is the root mean square of successive R–R interval differences (RMSSD), which reflects the beat-to-beat variance of HR and serves as a primary measure for estimating vagally mediated changes in HRV.49 Spectral-domain measures of HRV are the total power (TP, 0.0033–0.4 Hz), the high-frequency power (HF, 0.15–0.4 Hz), low-frequency power to TP (LF/TP, 0.04–0.15 Hz), high-frequency power to TP (HF/TP, 0.15–0.4 Hz) and low-frequency to high-frequency power (LF/HF). The ratio may provide an estimate of the balance between sympathetic nervous system (SNS) and PNS activity.

Statistical analysis

The critical level of significance for all statistical analyses will be set to p value <0.05. The analyses will be carried out using the Matlab and R statistical software packages. Missing data will be addressed using multiple imputation methods to minimise bias, ensuring robustness of the results.

Analysis of main endpoint indicators

Prior to conducting inferential statistics, Kolmogorov-Smirnov (K-S) tests will be employed to assess the normality of the measurement data. In cases where the data deviate from a normal distribution, transformation methods will be applied. Specifically, logarithmic transformation may be employed for data exhibiting extreme skewness, while the square root transformation can be adopted for data displaying moderate or small skewness. The examination of the difference in age, headache types and gender between the active taVNS and sham group will be conducted for the AP study and the PP study, respectively.

To evaluate the primary efficacy of the AP study (eg, VAS scores), a paired samples t-test will be performed before and after the treatment sessions for both the taVNS and sham group. This analysis will determine if there is a significant improvement in relief of headache after taVNS treatment or sham intervention. The VAS scores before and after the treatment sessions will be subtracted to calculate the improvement scores, representing the treatment effect. Independent samples t-tests will be conducted between the taVNS group and the sham group to assess the potential benefits of taVNS. The effect sizes of the treatment effects (Cohen’s d) will be documented.

Primary measures of the PP study between taVNS and sham groups will be examined using mixed analysis of variance (Mixed ANOVA), which is a statistical test used to analyse data that involve both between-subjects and within-subjects factors. The headache attack days of the 4 week baseline, the last 4 week of intervention and the last 4 week of open label periods of two groups will be tested by mixed ANOVA. Pairwise comparisons will be performed for three sets of data for each group, respectively. To address the risk of alpha inflation, Bonferroni correction will be applied to counteract the problem of multiple comparisons.

Analysis of secondary endpoint indicators

Secondary outcome measures between groups will be statistically examined by mixed ANOVA, paired t-test and independent samples t-test to examine the efficacy of taVNS. The statistical results will be corrected based on Bonferroni’s principle as this statistical operation requires multiple comparisons, and there will be a bias of Alpha inflation. In addition, correlation analysis will be conducted between the change in VAS scores, the change in headache attack days (ie, treatment efficacy index) and the HRV features.

For the analysis of the EMG features, the subjects in each group will be further divided into three subgroups based on their diagnosis, that is, migraine, CH and TTH. To investigate whether EMG features can be used to distinguish three types of headaches, ANOVA will be performed to test the difference in the EMG features of these three groups before the treatment in each study. For two TTH subgroups (receiving taVNS vs receiving sham), mixed ANOVAs will be performed on EMG features from pretest and post-test to examine the treatment efficacy.

Records of side effects

TaVNS does not have any known long-term risks, and no serious side effects or safety accidents have been mentioned in the existing studies. Participants in the process of using the instrument may experience some, all or none of these adverse reactions: skin pain in the ear, tinnitus, hoarse throat, nausea and vomiting. When the above symptoms occur, the subject should immediately communicate with the physician and discontinue the intervention. In most cases, the symptoms will disappear within a period of time after stopping the intervention.