Obesity is a major public health concern worldwide, with a growing number of individuals affected. Despite the availability of various weight loss interventions, many individuals struggle to achieve and maintain a healthy weight. One treatment for obesity is AS, which has been shown in clinical trials to be effective in reducing stress and anxiety (Usichenko et al., 2022). Embedded ear needles were popularized in the ‘70s and ‘80s for weight loss (Schatz, 1975), which might have propagated the conduct of many clinical trials using auricular therapy for various conditions (insomnia, cocaine addiction, back pain, and epilepsy; Gates et al., 2006; Lan et al., 2015; Liu et al., 2018; Moura et al., 2019; Mendonça et al., 2020; Yap et al., 2020). Currently, it is presumed that AS exerts its effects through the involvement of cranial nerves V, VII, and X, which lead 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). The main response is seen in connection to the vagal nerve, which presents on the auricle (Arnold’s nerve) and can be directly stimulated in the cymba and the region around the meatus acusticus externus (Peuker and Filler, 2002; Peuker, 2003). This study aimed to provide a comprehensive and systematic review of RCTs assessing the effects of AS on weight reduction and obesity-related parameters such as waist circumference, waist-to-hip ratio, and blood lipids in obese patients. The study aimed to provide a better understanding of the potential benefits and limitations of AS as a treatment strategy for weight loss in obesity and its effect on obesity-related parameters.

Systematic review and meta-analysis to evaluate the effects of AS on obesity-related parameters and the safety of AS.

Fixed-effect meta-analyses were performed initially, in cases of high heterogeneity, random-effect analyses were performed. Data not suitable for meta-analysis were reported separately. Furthermore, for each anthropometric parameter, a GRADE assessment was undertaken.

Out of 1,274 trials that were analyzed with full-text analysis, a total of 22 met the inclusion criteria (see Figure 1).

The number of trial participants, gender, age, type of intervention, and assessment method for anthropometric parameters are summarized in Table 2.

A total of 1,333 individuals were included in this systematic review. A total of 91% were female subjects. One trial did not provide any information on sex concerning the included school children (Cha and Park, 2020). The age of the individuals ranged between 10 and 62 years, with a median of 35 years. One trial examined a mixed population of obese and non-obese patients (Yeh and Yeh, 2008). Obesity was defined as a BMI >25. Two studies adjusted the cutoff for obesity because of the Asian population (Hsieh, 2007, 2010). In the subgroup analysis, we pooled only the data of the obese trial participants. One trial was conducted on psychiatric in-patients (Ching et al., 2012), and another on a sample with knee osteoarthritis (Costa-Cavalcanti et al., 2018).

Several study designs could be identified. Most of the studies were designed as two-armed studies (n = 16) (Allison et al., 1995; Hsieh, 2007; Yeh and Yeh, 2008; Hsu et al., 2009; Shen et al., 2009; Hsieh et al., 2011; Ching et al., 2012; Darbandi et al., 2012; He et al., 2012; Kim et al., 2014; Schukro et al., 2014; Yeh et al., 2015; Cayir et al., 2017; Costa-Cavalcanti et al., 2018; Lillingston et al., 2019; Cha and Park, 2020), followed by three-armed studies (n = 5) (Mok et al., 1976; Qunli and Zhicheng, 2005; Hsieh, 2010; Lien et al., 2012; Yeo et al., 2014) and four-armed studies (n = 1) (Darbandi et al., 2014). All the included trials were conducted on obese individuals.

Thirteen RCTs used acupressure with small beads or plant seeds for stimulation (Allison et al., 1995; Hsieh, 2007, 2010; Yeh and Yeh, 2008; Hsieh et al., 2011; Ching et al., 2012; Darbandi et al., 2012, 2014; He et al., 2012; Kim et al., 2014; Yeh et al., 2015; Costa-Cavalcanti et al., 2018; Cha and Park, 2020). Auricular acupuncture with conventional needles or semipermanent needles was identified in eight studies (Mok et al., 1976; Qunli and Zhicheng, 2005; Hsu et al., 2009; Shen et al., 2009; Lien et al., 2012; Schukro et al., 2014; Yeo et al., 2014; Cayir et al., 2017; Lillingston et al., 2019). Additional electrical stimulation of the ear was used in one trial (Schukro et al., 2014). In 21 of all 22 trials, stimulation was applied in auricular regions with vagal innervation (mainly concha). One study did not provide any information about the location of AS (Hsieh, 2010).

Eleven trials used add-on AS to the use of additional weight loss interventions, such as exercise, diet, or weight reduction education, which included diet and physical exercise counseling for all treatment arms (Hsieh, 2007, 2010; Hsieh et al., 2011; Darbandi et al., 2012; Schukro et al., 2014; Yeo et al., 2014; Yeh et al., 2015). The mean duration of AS was 48 days, ranging between 21 days (Mok et al., 1976) and 87 days (Cayir et al., 2017).

Different control groups were identified in the included studies. The most common control group was a sham procedure (Mok et al., 1976; Hsieh, 2007, 2010; Hsu et al., 2009; Shen et al., 2009; Hsieh et al., 2011; Ching et al., 2012; Darbandi et al., 2012, 2014; Lien et al., 2012; Schukro et al., 2014; Yeo et al., 2014; Yeh et al., 2015; Costa-Cavalcanti et al., 2018; Lillingston et al., 2019; Cha and Park, 2020). Some studies have also compared AS vs. active controls (Mok et al., 1976; Hsieh, 2007, 2010; He et al., 2012; Lien et al., 2012; Wang et al., 2014; Yeo et al., 2014). These included mostly dietary instructions (Hsieh, 2007, 2010; Yeo et al., 2014) and physical exercise (He et al., 2012). Four studies also compared AS to classical body acupuncture and acupressure (Allison et al., 1995; Qunli and Zhicheng, 2005; Darbandi et al., 2014; Cayir et al., 2017). Four trials had no-intervention control (Yeh and Yeh, 2008; Kim et al., 2014).

All trials applied AS on regions of vagal innervation via the auricular branch of the vagus nerve (ABVN) (see Table 1 and for most frequently used points, see Figure 2). Most trials have examined the effects of AS on anthropometric parameters. Weight (n = 17) (Mok et al., 1976; Allison et al., 1995; Qunli and Zhicheng, 2005; Yeh and Yeh, 2008; Hsu et al., 2009; Shen et al., 2009; Hsieh et al., 2011; Ching et al., 2012; Darbandi et al., 2012; He et al., 2012; Lien et al., 2012; Kim et al., 2014; Schukro et al., 2014; Yeo et al., 2014; Cayir et al., 2017; Lillingston et al., 2019; Cha and Park, 2020), BMI (n = 12) (Hsieh, 2007, 2010; Yeh and Yeh, 2008; Hsu et al., 2009; Darbandi et al., 2012, 2014; He et al., 2012; Lien et al., 2012; Kim et al., 2014; Schukro et al., 2014; Yeo et al., 2014; Cayir et al., 2017; Lillingston et al., 2019), body fat (n = 6) (Allison et al., 1995; Yeh and Yeh, 2008; Darbandi et al., 2012; Kim et al., 2014; Yeo et al., 2014; Cayir et al., 2017), waist circumference (n = 6) (Yeh and Yeh, 2008; He et al., 2012; Lien et al., 2012; Darbandi et al., 2014; Yeo et al., 2014; Cayir et al., 2017), hip circumference (n = 4) (Yeh and Yeh, 2008; Lien et al., 2012; Darbandi et al., 2014; Cayir et al., 2017), and the ratio of waist and hip circumference (n = 2) (Yeh and Yeh, 2008; Kim et al., 2014). Three trials examined the effect on glucose metabolism (Hsu et al., 2009; Lien et al., 2012; Costa-Cavalcanti et al., 2018), and a total of five studies have investigated the effects related to lipid metabolism and digestive hormones (Hsu et al., 2009; Darbandi et al., 2012; Lien et al., 2012; Yeh et al., 2015; Costa-Cavalcanti et al., 2018). Of the 22 trials included in this systematic review, 15 provided sufficient data and were eligible for meta-analysis, while 7 trials were evaluated descriptively. Forest plots are only presented for pooled outcomes of five trials and more. Forest plots of four or fewer trials can be found in the Supplementary material.

A total of 10 trials with 487 individuals provided data on BMI. Compared to the control methods, AS significantly reduced BMI (MD = −0.38 BMI points, 95% CI (−0.55 to −0.22), p < 0.0001) (see Figure 3). The strongest BMI reduction of a mean of 1.01 points was achieved in a two-armed trial by Kim et al. (2014). South Korean female obese college students were allocated to ear acupressure or waiting lists. After 1 month of continuous and self-applied ear acupressure with diet or exercise, the acupressure group lost 3.1 kg (SD 0.73) vs. 0.2 kg (SD 1.05) (difference p < 0.001) and 1.23 (SD 0.34) BMI points vs. 0.15 (SD 0.45) BMI points (difference p < 0.001) compared to the control group in the pre-post calculation. The authors explained that satisfying results were connected to self-treatment and an increased feeling of self-efficacy. A similar reduction of weight and BMI was achieved in the sham-controlled trial of Lien et al. Taiwanese obese women received auricular true or sham acupuncture. Regular diets were maintained throughout the study. After 1 month of ear acupuncture with semipermanent needles, the real acupuncture group had a mean of 1.3 kg (SD 2.2) vs. 0.6 kg (SD 1.4) and a BMI of 0.5 points (SD 0.9) vs. 0.2 points (SD 0.5) in the sham acupuncture group.

In 12 trials with data on the body weight of 655 individuals, compared to the control methods, AS led to a significant reduction of body weight (MD = −0.66 kg, 95% CI (−1.12 to −0.20), p = 0.005) (see Figure 4).

No significant difference, despite a low-calorie diet, could be achieved with add-on acupressure compared to add-on sham acupressure in Iranian obese men after 6 weeks of treatment (Darbandi et al., 2014).

Data of 6 trials with 317 individuals were included in the analysis of body fat. In body fat percentage, compared to the control methods, AS did not show a significant effect on reducing the amount of body fat (MD = −1.21, 95% CI (−2.47 to 0.05), p = 0.06) (see Figure 5). Body fat was mostly measured with the method of impedance analysis. Cayir et al yielded the highest body fat percentage reduction from a baseline of 5.6% (SD5.3) with auricular acupuncture vs. 2.0% (SD2.8) with body acupuncture. No additional measures were used by the patients, and both acupuncture forms were applied two times per week over 12 weeks.

In the analysis of the six trials with data on waist circumference of 276 individuals, compared to the control methods, AS reduced waist circumference significantly (MD = −1.44 cm, 95% CI (−2.69 to −0.20), p = 0.02) (see Figure 6). The biggest decrease was achieved in the trial by Darbandi et al. (2014), who randomized 80 obese participants in four groups: body electroacupuncture (A), auricular acupressure (C), sham body electroacupuncture (B), and sham auricular acupressure (D). From the auricular acupressure vs. sham auricular acupressure arms (n = 2×20), it was possible to use the data for meta-analysis. All subjects received a 500 kcal low-calorie diet. Only the verum auricular acupressure led to a relevant reduction of hip and waist circumference after 6 weeks of treatment compared to sham (Darbandi et al., 2014).

Regarding the pooled data from four or fewer trials (see forest plots in the Supplementary material), a positive effect on obesity-related metabolic parameters was observed only in the comparisons of leptin, insulin, and HOMA insulin resistance. Compared to the control methods, AS had a significant effect on reducing leptin (4 trials, n = 295, SMD = −0.40, 95% CI (−0.63 to −0.17), p = 0.0008). Regarding insulin, AS had a significant effect on reducing fasting insulin serum level (3 trials, n = 139, MD = −3.50 UI/ml, 95% CI (−6.59 to −0.41), p = 0.03) and HOMA insulin resistance (3 trials, n = 139, MD = −1.06, 95% CI (−2.03 to −0.09), p = 0.03).

No significant effects of AS compared to control methods were found in the meta-analyses of hip circumference, ratio of waist/hip circumference, cholesterol, LDL, triglycerides, adiponectin, and ghrelin (see Supplementary material). Only one trial examined the impact of glucose levels in obese patients (Costa-Cavalcanti et al., 2018). No significant difference was reported between the groups.

For details, see Tables 3, 4. We performed two sensitivity analyses: First, positive outliers were excluded from the meta-analysis.

Second, trials with additional interventions besides AS were excluded. In the case of waist circumference and leptin, significant results could be still obtained, excluding positive outliers.

Regarding the sensitivity analysis of AS, excluding additional interventions such as dietary advice or physical exercise, we still found a significant benefit of AS alone for reducing BMI and weight.

There were seven trials that did not have continuous data for quantitative meta-analysis. In Table 5, the dichotomous outcomes are summarized.

Of the 22 included studies, adverse events were recorded in 12 studies. Six study groups could not document any adverse side effects (Mok et al., 1976; Yeh and Yeh, 2008; Darbandi et al., 2012, 2014; Cayir et al., 2017; Cha and Park, 2020). Minor side effects were recorded in five trials (Allison et al., 1995; Hsu et al., 2009; Ching et al., 2012; Lien et al., 2012; Schukro et al., 2014). One study described in its methodology that they recorded adverse side effects. However, no information could be extracted from the results (Yeh et al., 2015).

Common side effects of AS that have been reported are all mild and include pain or minor bleeding at the stimulation side (four trials), transient dizziness (three trials), erythema (two trials), and skin irritation (two trials).

The overall study quality was moderate (see Figures 7, 8). A funnel plot was performed and revealed no publication bias.

One trial had several contradictions in the tables, and we judged it prudent to exclude it from the evaluation (Abdi et al., 2012).

Most study groups had not provided funding information (Mok et al., 1976; Qunli and Zhicheng, 2005; Hsieh, 2007, 2010; Shen et al., 2009; Hsieh et al., 2011; Lien et al., 2012; Kim et al., 2014; Schukro et al., 2014). For the most part, the studies were supported either by the research group itself (Hsu et al., 2009; Darbandi et al., 2012, 2014; Yeh et al., 2015; Cayir et al., 2017; Lillingston et al., 2019) or by the government (Yeh and Yeh, 2008; Ching et al., 2012; Yeo et al., 2014; Costa-Cavalcanti et al., 2018; Cha and Park, 2020). Individual studies were funded by a company (Allison et al., 1995) or an acupuncture society (He et al., 2012).