Cardiovascular effects of auricular stimulation -a systematic review and meta-analysis of randomized controlled clinical trials

Background The number of randomized controlled trials using auricular stimulation (AS) such as transauricular vagus nerve stimulation, or other auricular electrostimulation or auricular acupuncture or acupressure, in experimental and clinical settings, has increased markedly over the last three decades. This systematic review focusses on cardiovascular effects of auricular stimulation. Methods and analysis The following databases were searched: MEDLINE (PubMed), EMBASE, Cochrane Central Register of Controlled Trials (CENTRAL), ISI Web of Science, and Scopus Database. RCTs were reviewed that had been published in English and European languages. Data collection and analysis was conducted by two reviewers independently. Quality and risk assessment of included studies was performed and the meta-analysis of the effect of the most frequently assessed biomarkers. Results Altogether, 78 trials were included. 38 studies assessed heart rate (HR), 19 studies analyzed heart rate variability (HRV), 31 studies analyzed blood pressure (BP) and 7 studies were identified that measured oxygen saturation (O2), 2 studies on baroreflex sensitivity and 2 studies on skin conductance were evaluated in this review. 26 studies contained continuous data and were eligible for meta-analysis, 50 trials reported non continuous data and were evaluated descriptively. The overall quality of the studies was moderate to low. AS leads to a significant reduction of HR, the changes though were not considered an adverse reaction. Furthermore, when looking at HRV, AS was able to reduce the LF/HF ratio significantly compared to control procedures. No other cardiovascular parameters (blood pressure, oxygen saturation, baroreflex sensitivity) were changed significantly. AS produced only minor side effects in all trials. Conclusion AS can lead to clinically safe reduction of HR and changes in the LF/HF ratio of the HRV, which is presumably via an increase in vagal activity. More research is needed to clarify whether AS can be used to modulate tachycardia or indications with autonomic imbalance. Systematic review registration https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=231885 PROSPERO, ID CRD42021231885.


Introduction
Research evaluating the effectiveness of stimulation of the pinna (auricular stimulation, AS) has markedly increased in the last 30 years.AS trials apply a variety of techniques such as acupuncture or acupressure, or electrostimulation on distinct anatomical regions of the auricle.For an extended list of AS techniques see Table 1.The potential mechanism of AS effects is attributed to the neuroanatomical conditions of the external auricle.It is presumed that AS exerts its effects via the involvement of cranial nerves V, VII and X (Peuker and Filler, 2002).Alderman's nerve or Arnold's nerve is a branch of the vagus nerve which forms a receptive field in the pinna of the ear; stimulation in these areas are thought to lead to vagal activation and to the modulation of brain areas involved in stress response, such as the limbic system, locus coeruleus and hypothalamus (Qu et al., 2014;Frangos et al., 2015).Smoking cessation, drug withdrawal, pain relief, heart rhythm disorders, epilepsy, insomnia and depression, and obesity treatments are among the most frequently evaluated conditions (Gates et al., 2006;Lan et al., 2015;Liu et al., 2018;Moura et al., 2019;Mendonça et al., 2020;Yap et al., 2020).Biomarkers, such as blood samples measuring metabolic profiles, inflammatory or immunological markers, anthropometric data such as weight, BMI, as well as cardiological and neurological electrophysiological measurements and functional neuroimaging are used as objective outcomes.
However, an extensive overview including all types of AS and their effects on cardiovascular biomarkers, is missing.This systematic review was developed following the PRISMA guidelines to explore and evaluate-to our knowledge for the first time-the existing literature regarding the effect on cardiovascular parameters in randomized controlled trials comparing AS with sham AS or AS with no intervention.This review also aimed to investigate whether systemic effects from AS are clinically significant and helps to identify the potential for future clinical research for AS.

Objectives
The aim of this systematic review and meta-analysis is to evaluate the effects of AS on cardiovascular parameters and the safety of AS in healthy individuals and patients.

Methods
The systematic review protocol has been registered on PROSPERO (ID CRD42020184795).Since all data used in this systematic review have been published, this review does not require ethical approval.
2.1.Eligibility criteria for included trials in the review.

Types of trials
The review included only randomized controlled trials (RCTs) in English and other European languages.The funding source was registered.Systematic review and meta-analysis were conducted according to the Cochrane Handbook for Systematic Reviews of Interventions.

Types of participants
The review comprised randomized controlled studies in clinical settings (with patients) and experimental settings with healthy individuals.No restrictions regarding age, gender, or ethnicity of health conditions were made.

Types of interventions
We included all RCTs applying auricular stimulation alone or in addition to routine care.All interventions were eligible from traditional AS (i.e., auricular acupuncture, auricular electroacupuncture, auricular acupressure) to related techniques such as the electrical transcutaneous auricular vagus nerve stimulation (tVNS) in the conchae of the auricle or cranial electrotherapy stimulation (CES) with electrodes clipped to each earlobe.Any comparison with control conditions (sham acupuncture, sham  acupressure, placebo stimulation, routine care etc.) were included.We excluded trials that compared one AS to another AS technique.

Types of outcome measures
2. 1.4.1. Outcomes We screened for the following parameters as part of our eligibility review of studies: heart rate (HR), systolic blood pressure (SBP), diastolic blood pressure (DBP), Heart rate variability (HRV,), Low frequency (LF), High frequency (HF), LF/HF ratio und oxygen partial pressure.All cardiovascular biomarkers, that were reported with results were extracted and evaluated, non-continuous data were extracted and evaluated separately.In serial measurements we chose the values at the end of the intervention period.Adverse event reporting was analyzed.

Search methods for identification of trials
2.2.1.Electronic searches.

Trial identification
Two researchers (JD, KH) screened independently: titles, abstracts, and full texts for eligibility.Disagreements were resolved by discussion with a third author (TU).If an article did not provide enough information to decide about eligibility, we contacted the trial authors via e-mail.The selection process is depicted in the PRISMA Flow Chart in Figure 1 (Page et al., 2021).Management of selected studies was done with the help of the Covidence software (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia 1 ).

Data extraction and assessment of risk of bias in included trials.
JD and KH independently extracted data and evaluated the methodological quality of each RCT using Risk of Bias domains recommended in the Cochrane Handbook following the RoB 1 tool 1 Available at www.covidence.org with some modifications.A consensus procedure was performed before entering the data into Review Manager software (RevMan 5.4.2020).

Measures of treatment effects and dealing with missing data
For non-continuous outcomes the effect measures of choice were analysed descriptively.Cardiovascular outcomes, that were presented as continuous data were analysed as mean differences with 95% confidence intervals (CI), or as standardized mean differences (SMD).If final means were not reported, we used changes from baseline in the meta-analysis.As well, in case of very different baseline values between groups, changes from baseline were used in the final calculation instead.If relevant numbers of data were missing, we reported it in the Risk of Bias section.We did not apply imputation or other strategies for missing data.

Assessment of heterogeneity
Owing to the broad inclusion criteria a high heterogeneity was to be expected.We only applied random effects meta-analysis (RE), instead of fixed effect meta-analysis (FE) where we considered high heterogeneity a relevant issue.
Heterogeneity was regarded substantial if T 2 is greater than zero and either I 2 is greater than 50% or in case of low p value (less than 0.10) in the Chi 2 test for heterogeneity.The measure T 2 can be calculated directly from Cochran's Q.Here, the individual deviations are weighed according to the precision of the respective individual studies, i.e., studies with low random scatter are considered more than studies with lower precision and thus have less influence on the estimation of the meta-estimate.I 2 according to Higgins/ Thompson: I 2 can be interpreted as the ratio of the variance between the studies to the total variance in the meta-analysis.

Assessment of reporting biases
A funnel plot with asymmetry was examined for each of the analyses and is provided in the Supplementary material.

Data synthesis
Fixed-effect meta-analysis were performed initially, in cases of high heterogeneity, random effect analyses were performed.Dichotomous data were analysed separately.The "fixed effects" approach assumes that effects are constant across studies, therefore it is only appropriate when heterogeneity between studies is negligible.In contrast, the "random effects" approach both within-study variance and between-study variance are considered to estimate the aggregate effect.The "random effects" approach is appropriate when heterogeneity between studies is significant and studies are expected to measure different true effects (Borenstein et al., 2010).
A Grading of Recommendations, Assessment, Development and Evaluation (GRADE) was performed for each cardiovascular parameter.With GRADE, the evidence is summarized in a summaryof-findings table, rating the certainty of evidence and the relative and absolute treatment effects for each important endpoint.
Subgroup analysis was performed to assess the effects of the different auricular stimulation methods.If the subgroup analysis using FE demonstrated heterogeneity, the data analysis was performed with the RE model.If we detected a wide dispersion of results between studies or limitation of the quality of the results due to methodologically weak studies, a sensitivity analysis was performed excluding outliers or limited to studies with low risk of bias.

Literature search and analysis
Out of 1.274 trials that were analyzed with full-text analysis a total of 78 trials contained outcomes with cardiovascular parameters.

Data extraction and analysis
The number of trial participants, gender, age, type of intervention and assessment method for cardiovascular parameter are summarized in Table 2 with the overview of the included studies.

Quality assessment
The analysis of study quality was performed for the included studies in the meta-analysis.The overall study quality was moderate (see Figures 2, 3).Missing data did not bias the review findings in general.

Baseline characteristics
In total 3,777 patients were included in the systematic review of cardiovascular parameters.55,5% were female.Four studies reported no information of the sex of the included population (Taylor and Lee, 1992;Strong et al., 2016;Abdi et al., 2017;Dellovo et al., 2019).The age of the patients ranged between 4 days and 86 years, with a median of 30 years.One study did not provide information about age (Strong et al., 2016).
Forty-seven studies used tVNS or CES as the stimulation variant.Interventions such as tVNS or CES apply electrical current to broader areas of the auricle such as the cymba concha or the tragus in tVNS or the earlobes such as in CES.According to Peuker and Filler (2002), the cymba conchae is innervated by the auricular branch of the nervus vagus (ABNV), the tragus has a mixed innervation of ABNV plus great auricular nerve (GAN, a superficial branch of the cervical plexus) and the earlobe is innervated primarily by the GAN and is therefore frequently used as sham-control.For the most frequently used areas of stimulation in the 78 included trials see Figure 4.

Outcomes and safety parameters
Thirty-eight studies assessed heart rate (HR), 19 studies analyzed heart rate variability (HRV), 31 studies analyzed blood pressure (BP) and 7 studies were identified that measured oxygen saturation.In addition, 2 studies on baroreflex and 2 studies skin conductance were evaluated in this review.
Of the 76 studies identified, 26 studies contained continuous data and were eligible for meta-analysis, 50 trials reported non continuous data and were evaluated descriptively (see Table 3).

Heart rate
Seventeen studies (Taylor and Lee, 1992;Wang et al., 2004;Barker et al., 2006;Lin et al., 2011;Busch et al., 2013;Yeh et al., 2015;Klausenitz et al., 2016;Kuo et al., 2016;Antonino et al., 2017;Chen et al., 2017;Sabino-Carvalho et al., 2017;Fischer et al., 2018;Ventura-Bort et al., 2018;Villani et al., 2019;Gan et al., 2020;Giraudier et al., 2020;Ricci et al., 2020;Vosseler et al., 2020) had data regarding HR (see Figure 7).The results suggest that auricular stimulation significantly reduces heart rate compared with control procedures MD = −1,23, 95% CI (−1.74 to −0.72), p = 0.0005.No drop of HR in any of the trials were regarded as adverse event.The strongest decrease was achieved in the auricular acupressure subgroup.Here Kuo et al. (2016) conducted a trial with 80 postpartum women, with the primary outcome to relieve stress and anxiety.The intervention group received auricular acupressure together with routine care over 4 days compared to a control group that received routine care only.The acupressure was administered by a researcher.HR was measured as secondary outcome parameter and showed to be lower in the intervention group compared to the routine care group by a mean of 9.2 bpm [CI95% 13.27, 5.13].Barker et al. (2006) conducted a trial in 38 elder patients with fresh hip fracture to reduce pain and anxiety on the way to the hospital.Acupressure was administered by the paramedics before the transfer and compared to sham acupressure.The intervention group had significantly lower pain and anxiety and heart rates arriving at the hospital by a mean of 18 bpm.Interestingly this trial was a doubleblind trial since the paramedics did not know if they administered at real or sham points on the auricle.
The reduction of HR in the electrostimulation group was statistically significant, but not clinically relevant-the strongest reduction was achieved in the experiment by Taylor and Lee (1992) on 90 healthy volunteer students, who received in a 5-armed randomized trial design 30 min of TENS electrostimulation of different intensities (0-5 kHz, 0-10 mA) to both earlobes or no stimulation.This safety trial monitored HR and blood pressure, and

Outcome and no. of RCTs Only in AS group No difference between AS and controls
Low frequency power N = 2 Increase: 1 (Stavrakis et al., 2020) Reduction: 1 (Zhu et al., 2021) 0 Oxygen Saturation N = 3 0 3 (Karst et al., 2007;Strong et al., 2016;Dellovo et al., 2019) anxiety.The strongest reductions of HR were achieved in the 100 Hz group resulting in a mean 89.22 bpm + − SD 6.13 vs. 101.71bpm + − SD 7.92 in the placebo TENS group.

Sensitivity analysis
Although the outlier that was excluded in favor of auricular stimulation in the sensitivity analysis, a significant reduction in heart rate could still be detected: MD (95%CI)-1.09[−1.61, −0.58], p 0.0001.

FIGURE 6
Diastolic blood pressure: auricular stimulation vs control.Heart rate: auricular stimulation vs control.Heart rate variability: auricular stimulation vs control.
studies demonstrated, that auricular stimulation had a significant effect over control procedures with lower ratios in the AS group (MD = −0.14, 95% CI (−0.23 to 0.04), p = 0.007) (see Figure 11).In order to measure the effects of the electrical stimulation of the ABVN on the autonomic nervous system, Clancy and her colleagues (2014) (Clancy et al., 2014) conducted an experimental RCT in 48 healthy participants.34 received 15 min of 30 Hz tVNS with an intensity to the level of sensory threshold on the inner and outer surface of the tragus, compared to a sham group with inactivated electrodes on the tragus.HRV was measured and a significant decrease in LF/HF ratio during active tVNS could be shown in comparison to the sham group that did not show a significant decrease.
Another experimental cross-over RCT on 15 healthy men by Vosseler et al. (2020), measured the effects every 30 min in 120 min of tVNS with a frequency of 25 Hz and 2.5 mA on the cymba conchae, compared to the earlobe in the sham group.The earlobe is supposed to be free of vagal nerve fibers.Endocrine and metabolic parameters and on peripheral vagal activity during an oral glucose tolerance test were the outcome parameters.Significant reduction in LF/HF ratio were achieved at the end of the stimulation (at 120 min).
Heart rate variability (low frequency): auricular stimulation vs control.

FIGURE 11
Low-frequency/high-frequency ratio (LF/HF): auricular stimulation vs control.Oxygen saturation: auricular stimulation vs control.Antonino et al. (2017) tested the hypothesis that tVNS acutely improves spontaneous cardiac baroreflex sensitivity (cBRS) and autonomic modulation in a cross over RCT in 13 healthy men.Bilateral tVNS on the tragus over 15 min with 30 Hz between 10 and 50 mA up to the level of sensory threshold was compared to sham tVNS with no current and tVNS on the earlobe.HRV was measured during the stimulation and tVNS significantly reduced LF/HF ratio and returned to baseline values during recovery.

Oxygen saturation
Continuous data of O2 saturation were provided in 3 trials (WANG et al., 2009;Chen et al., 2017;Gan et al., 2020).In 2 trials (Chen et al., 2017;Gan et al., 2020) they were part of secondary parameters to objectify indirectly the intensity of pain (intense crying) in neonates undergoing unpleasant examination procedures.In both trials acupressure led to significantly lower pain perception, though only in the trial by Gan et al. (2020) on 100 neonates undergoing painful eye inspection, AS prevented significantly from intense crying with subsequent O2 reduction.The trial by WANG et al. (2009) was conducted in 45 adults with Obstructive Sleep Apnoe Syndrome (OSAS) and impaired O2 saturation at night-a 3 month lasting acupressure led to relevant improvement of sleeping patterns and a significant improvement of nocturnal O2 levels after the end of treatment (see Figure 12).

Baroreflex sensitivity and skin conductance
AS led to increased baroreflex sensitivity in two studies with tVNS (see Supplementary Figure S2); two studies showed no difference between groups in Skin conductance MD = −0.47,95% CI (−1.00 to 0.05), p = 0.08 (see Supplementary Figure S3).

GRADE assessment
The overall GRADE assessment for the outcomes systolic and diastolic blood pressure, heart rate, heart rate variability was performed using the GRADE tool.Recommendations of AS for reduction of heart rate were low and very low in the other outcomes (see Supplementary table S1).

Discussion
In the present systematic review, we have screened various parameters as potential biomarkers for the effects of auricular stimulation on cardiovascular function of the human body.Although these parameters were not the primary outcomes in the RCTs included in this review, we have found a clinically significant reduction of HR as well as reduced LF/HF ratio after auricular stimulation compared to control procedures.This finding sounds physiologically plausible, since the main mechanism of auricular stimulation suggested is the modulation of autonomic nervous system (Usichenko et al., 2017) and Hua et al. 10.3389/fnins.2023.1227858Frontiers in Neuroscience 15 frontiersin.orgboth effects found (reduction of HF and LF/HF ratio) are consistent with the physiological reaction to the stimulation of the auricular branch of the vagal nerve (Peuker and Filler, 2002).
No other cardiovascular parameters (blood pressure, oxygen saturation, baroreflex sensitivity) were changed significantly.
The electrical stimulation of the parasympathetic nervous system via the auricular branch of the vagus nerve (Alderman's nerve or Arnold's nerve) (Peuker and Filler, 2002), raised concerns about cardiovascular safety of AS, especially in an elderly population with cardiovascular comorbidities.These concerns are based on experimental investigations with direct stimulation of cervical vagal nerves in dogs that demonstrated the more pronounced effect of right vagal nerve stimulation on bradycardia in comparison with the left side stimulation (Ardell and Randall, 1986).However, the simple theoretical transfer of this effect in case of AS is not justified, since the neurocircuitry of transcutaneous vagal nerve stimulation differs from the direct form, where the left-sided AS is suggested as equally safe as the right-sided AS (Chen et al., 2015).Moreover, it is well known, that especially in patients with heart diseases the sympathetic part of autonomous nervous system is pathologically activated and is the target for various pharmacological interventions (e.g., block of betaadrenergic receptors) (de Lucia et al., 2019).Thus, in this cohort of patients, AS can serve as an additional non-pharmacologic method of myocardial protection due to reduction of heart rate and favorable modulation of the tone of autonomic nervous system.Indeed, tVNS is being investigated for modulating arrhythmias such as in atrial fibrillation (Yu et al., 2013;Stavrakis et al., 2015).It was demonstrated that tVNS relieved angina pectoris complaints, reduced heart rate and blood pressure and reduced the incidence of heart failure in comparison with the control group due to an inhibition of norepinephrine release from sympathetic nerves with subsequent dilation of cardiac microcirculatory vessels and improved left ventricular contractility in patients with severe coronary artery disease (Zamotrinsky et al., 2001).
One of the strengths of our work is that it has summarized the effects on cardiovascular factors of a broad range of AS, including traditional ways such as ear-acupuncture and modern ways such as electrical tVNS.
One explanation for the comparably larger effects in the acupressure groups (sysBP, HR) compared to the electrical stimulation such as in tVNS are the much longer duration of the stimulation in the acupressure group; here small beads or plant seeds are stuck with a tape to the auricle and remain in situ over several weeks, while the experimental trials with tVNS often apply electrical stimulation only maximum over several hours.The discrepancy of the results of the meta-analysis and the descriptive analysis of non-continuous data especially in the HR could be explained with the fact, that HR was used as a safety parameter in a large number of trials.Relevant changes were only reported if they were considered a safety issue, data of smaller changes were not reported.

Limitations
First of all the analysis was conducted with secondary outcome parameters, so conclusions are limited.Some of the trials with the strongest reduction in HR for example were investigating the effect of acupressure on anxiety and stress and pain as primary outcomes, therefore the cardiovascular effects could have been a secondary effect to the relief of anxiety and/ or pain (Barker et al., 2006;Kuo et al., 2016).Next, the literature search was limited to English and other European languages.A relevant number of studies is published in Chinese and could therefore not be evaluated.Furthermore, the meta-analysis offers limited power due to the moderate study quality of the included studies.In some subgroups the numbers of included studies of included studies are so small, that conclusions about effects cannot be drawn.

Conclusion
The findings of this systematic review support previously suggested mechanism of AS via activation of parasympathetic nervous system.The reduction of heart rate was clinically safe, the potential myocardial protective effect due to avoiding of tachycardias and modulating of autonomic imbalance should be clarified in future original investigations.
Photobiomodulation on auriculotherapy points Photoelectric stimulation of defined ear points Respiratory-gated auricular vagal afferent nerve stimulation (RAVANS) Staplepuncture surgical staple implanted in the concha of the ear Transcutaneous vagus nerve stimulation (tVNS) Vibrotactile treatment

FIGURE 3
FIGURE 3Risk of bias graph across all included studies.

FIGURE 4
FIGURE 4Innervation of the pinna and area of stimulation in the trials.Red: ABVN-auricle branch of the vagus nerve, blue: GAN-great auricular nerve.Yellow: ATN-auriculo-temporal nerve.Modified fromPeuker and Filler (2002).

TABLE 1
Examples of auricular stimulation described in clinical trials.

TABLE 2
Overview of included studies.

TABLE 3
Descriptive analysis of studies with non-continuous data.