Ear Acupressure for Allergic Rhinitis: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

Background The treatment effects and safety of ear acupressure (EAP) for patients with allergic rhinitis (AR) have yet to be clarified. Objective To evaluate the effects and safety of EAP in AR patients. Design Systematic review of published studies. Methods A total of 24 English and Chinese databases (PubMed, EMBASE (Excerpta Medical Database), Cochrane Central Register of Controlled Trials, CINAHL, Informit, ScienceDirect, LILACS (Latin American and Caribbean Health Sciences), ProQuest, AMED, Blackwell Synergy, PsycINFO, Panteleimon, AcuBriefs, KoreaMed, IndMed, Ingenta, mRCT, ISI Web of Knowledge, ERIC, VIP Information (http://www.cqvip.com), China National Knowledge Infrastructure (http://www.cnki.net), Cochrane Library, Chinese Cochrane Centre Controlled Trials Register Platform, and Wanfang Chinese Digital Periodical and Conference Database) were searched from their respective inceptions to August 2020 to collect randomized controlled trials of ear acupressure for allergic rhinitis. We performed literature inclusion, data extraction, and trial quality evaluations. Methodological quality was assessed according to the Cochrane Handbook. Revman5.3 was used for all analyses. Results A total of 203 trials were identified and eleven studies involved 1094 participants aged 3–70 years. EAP was better than control group interventions in terms of effectiveness (risk ratio (RR): 0.51; 95% confidence interval (CI): 0.36–0.70; P < 0.0001). EAP was superior to sham EAP in terms of improvement of the total nasal symptom score (RR: −0.50; 95% CI: −0.96–0.05; P = 0.03), sneezing score (RR: −0.36; 95% CI: −0.59–0.12; P = 0.003), global QoL score (RR: 0.42; 95% CI: 0.04–0.08; P = 0.03), and eye symptom score (RR: −0.36; 95% CI: −0.67–0.05; P = 0.02). Conclusions Despite the positive results, it is premature to confirm the efficacy of EAP for treating AR. More high-quality studies are needed to confirm safety and efficacy.


Introduction
Allergic rhinitis (AR) is a global health problem. It is a common symptomatic, inflammatory, and immunological disorder of the nasal mucosa, characterised by four classic symptoms: sneezing, nasal itching, airflow obstruction, and clear nasal discharge caused by IgE-mediated reactions [1]. AR can be subdivided into intermittent (four symptoms for <4 days per week or for <4 weeks) and persistent disease (four symptoms for >4 days per week or for >4 weeks) [2].
A self-reported epidemiologic study suggested that 10-30% of adults have AR and no fewer than 40% of children have AR [3]. In other words, more than 60 million people suffer from AR in the United States annually. A European epidemiologic study reported a prevalence rate of 25% [4].
Avoidance of exposure to specific allergens, patient education, pharmacological treatment, and immunotherapy form the current management approaches. Among these methods, medications are the most often selected strategy. A stepwise medical treatment protocol was proposed at the ARIA workshop report [28]. However, treatment based on guidelines is not effective in all patients [29]. Hence, many allergic rhinitis sufferers seek complementary and alternative medicine (CAM).
EAP is an alternative therapy in which magnetic bead or the seed of cowherb is attached to specifically stimulate points on the pinna. From the point of view of traditional Chinese medicine (TCM), all the major energy lines (meridians where acupuncture points are situated) are directly or indirectly connected to the ear. EAP was a shown to be effective for relief AR symptoms [30,31]. However, a previous meta-analysis suggested that the benefit of ear acupressure for symptomatic relief of allergic rhinitis is unknown [32].
In this previously published meta-analysis, the authors summarized the evidence of EAP on AR. However, this study made a mistake in literature, including a duplicate publication of Rao et al. (2005) and Rao and Han et al. (2006). In their results section, the authors stated that "ear acupressure was more effective than herbal medicine, as effective as body acupuncture or antihistamine for the shortterm effect, but it was more effective than antihistamine for the long-term effect." is statement may have been exaggerated, and the quality of evidence was low. In the present systematic review and meta-analysis, we provide an updated summary of evidence to evaluate the safety and efficacy of EAP for patients with AR.

Methods
Methodologic issues were resolved with guidance from the Cochrane Handbook for Systematic Reviews of Interventions [33].

Search Strategy.
A total of 24 English and Chinese databases were searched from their inceptions to August 18, 2020. ese were PubMed, EMBASE (Excerpta Medical Database), Cochrane Central Register of Controlled Trials, CINAHL, Informit, Science Direct, LILACS (Latin American and Caribbean Health Sciences), ProQuest, AMED, Blackwell Synergy, PsycINFO, Panteleimon, AcuBriefs, KoreaMed, IndMed, Ingenta, mRCT, ISI Web of Knowledge, ERIC, VIP Information (http://www.cqvip.com), China National Knowledge Infrastructure (http://www.cnki. net), Cochrane Library, Chinese Cochrane Centre Controlled Trials Register Platform, and Wanfang Chinese Digital Periodical and Conference Database. e Chinese Clinical Trial Registry Centre was also retrieved for ongoing trials. References of related identified publications were checked for additional trials, and we contacted authors by e-mail or telephone for additional data where necessary. roughout the search process, the following key words were used: the combination of allergic, rhinitis, rhinallergosis, AR, allergy, rhinitis, hay fever, ear, acupressure, acupuncture, auricular, acupoint, sticking, randomized controlled trial, RCT, and their synonyms. Two authors (SQL and QXZ) screened all citations independently. Table 1 displays the search strategy of the Cochrane Library.

Eligible Criteria.
Randomised controlled trials of EAP for AR were taken into account regardless of language or publication year. Patients presenting with seasonal AR or perennial AR of any age or gender were all included. We compared EAP with conventional therapy or Chinese herbal medicine formula or acupuncture or electroacupuncture or surgical therapy or placebo regimens studies.

Ineligible Criteria.
Observational studies, case reports, case series, letter, qualitative studies, and uncontrolled studies were excluded. Quasi-RCTs are not truly RCT. Including quasi-RCTs in the review may be detrimental to the power of conclusion. Hence, quasi-RCTs were also excluded. Allergic rhinitis merged with allergic asthma or allergic conjunctivitis and other allergic diseases were excluded. is was performed because targeted drug combination methods in these studies could not be used to compare effects.
EAP as intervention in the control group was excluded. When EAP was compared with other types of CHMFs or some other alternative therapy such as moxibustion therapy, nose massage, plaster therapy, or acupoint injection as the intervention treatment group, these were excluded due to the idea that combination therapy would disturb the efficacy summary. EAP as intervention in two groups using different auricular points was also excluded because these studies can be identified as explorations for the stimulation effect of various auricular ear points. Diagnostic criteria were required because accurate diagnosis is a prerequisite for targeted treatment.
All titles and abstracts of identified studies were initially scanned independently by our two authors (JZ and SQL). e full-text articles were obtained for further screening for inclusion in this review by these two authors when needed. A determination was then made as to whether the studies met our inclusion criteria. Any conflicts or disagreement were resolved by discussion.

Outcome Measures.
Trials were required to include as outcome measures either relief of symptoms of AR or evaluation of the efficacy of EAP in AR. e efficacy of EAP for AR was set as primary outcomes. Improvement in quality 2 Evidence-Based Complementary and Alternative Medicine of life, improvement of symptom scoring and other scale, and adverse events were set as secondary outcomes.

Methodological Quality Assessment, Data Extraction, and Data
Analysis. e risk of bias was assessed according to the Cochrane Handbook for Systematic Reviews of Interventions. e latest version of this tool was updated in March 2011, version 5.1.0 (http://www.handbook.cochrane.org/). Risk of bias items included the following: randomization sequence generation, allocation concealment, blinding of participants or healthcare providers, detection bias, incompleteness bias, reporting bias, and other biases. Raw data of all included studies containing the details of authors, the publication information, and design information of the original study were separately extracted by three authors (DL, TL, and QG).

Data
Analysis. Review Manager software version 5.3 was used to pool our data to perform the meta-analysis. Risk ratio (RR) was chosen for dichotomous data (efficacy, recurrence rate, and adverse events). Confidence interval (CI) was set at 95%, and P < 0.05 was defined as statistically significant. Cochrane X 2 and I 2 tests were used to investigate the heterogeneity of data. e statistical heterogeneity was considered substantial when P < 0.05 and I 2 > 50%, and then, a random effect model was used to calculate the effect size. If P > 0.05 and I 2 < 50%, then the studies included were homogeneous, and a fixed effect model was applied. Sensitivity analysis was conducted to test whether the resulting war was robust by excluding the study one by one and comparing the rest of the studies' effects with all the studies' total effects. We pooled trials when the intervention forms of those studies were adequately similar. Specific subgroups were analysed according to similar intervention forms or similar design. Funnel plots were generated for more than ten studies.

Results
We initially identified a total of 157 trials using the specific search strategy. No unpublished or ongoing studies were found. Sixty-nine duplicated texts were excluded. After reviewing titles, abstracts, and keywords, 52 studies were excluded for failure to conform to inclusion criteria. irtysix studies had initially appeared to meet our inclusion criteria. After the full texts were read, 25 were excluded, and t studies finally met our inclusion criteria. e study selection process is outlined in Figure 1.
Interventions in five Chinese studies [34,35,37,38,43] were EAP versus CT. Four studies [34,35,38,43] compared EAP to CT, including levocabastine, beclomethasone propionate nasal spray, astemizole, and cetirizine [37]. One study compared EAP combined with cetirizine to cetirizine alone [37]. Among these studies, Rao et al. included three comparisons, that is, the experiment group (ear-acupressure) and two control groups (acupuncture and cetirizine). In order for the reader to accurately identify intervention data, we extracted cetirizine control group data from Rao et al. (2005a).
ere are 245 participants in Claire et al. (2014). is was an international, multicentre, randomized, single-blind, sham-controlled trial. e intervention was EAP vs. sham EAP. Acupuncture therapy was the intervention of the control group in three studies [34,36,40], and electroacupuncture was the intervention of the control group in one study [39]. In the remaining two studies, CHMF was the intervention in the control group; these were Xinqin granules [41] and Shetizhiqiu decoction [43].
All studies provided the details of points used for ear acupressure. Among the total 19 ear points used in these 11 studies, Nei Bi (TG4) was used in all the studies. e lung (CO14), kidney (CO10), spleen (CO13), and adrenal gland (TG2p) were applied 10, 8, 7, and 6 times, respectively. We presented the frequency of ear point use in a radar plot ( Figure 2).

Reasons for Study Exclusion.
After our two authors read full texts of all included studies, we excluded 25 trials. e reasons for exclusion of those studies were as follows: (1) If a trial did not mention "diagnostic criteria," we excluded it on the basis of having "diagnostic criteria reason" (2) If a trial did not meet inclusion criteria for interventions, we excluded it on the basis of "intervention reason" (3) If a study was a repeated publication, we excluded it on the basis of being "duplication" Overall, there were ten "diagnostic criteria reason" studies, 14 "intervention reason" trials, and six that had both. ose were excluded for incomparability of the interventions. One "duplication" was identified. Characteristics of the 25 excluded studies are displayed in Table 3.
A list of 25 excluded studies by reading full text is displayed in Table 4.
In six RCTs [35-38, 40, 42], the method of randomization and allocation was not mentioned. Because of the lack of allocation concealment and the details of the randomization procedure, there might be a high risk of selection bias in these trials.
In the study by Lu et al. (2015) [39], randomization and the random allocation sequence concealment were reported. Hence, a low risk of bias could be defined in these trials.  [41] mentioned that a random number table tool was used for the allocation of participants. However, in the original texts, the authors failed to clearly describe the details regarding how the participants were allocated and concealed. erefore, there was a high risk of selection bias in these studies. , randomization numbers were reported as generated by an independent statistician using a computer system and sealed in individual opaque envelopes. In the study Lu et al. (2015) [39], to blind the outcome assessor, researchers who knew the whole course of treatment were not involved. In this way, the risk of performance and detection bias were be classified as low. e remaining 9 trials failed to mention their blinding [34][35][36][37][38][40][41][42]. Perhaps blinding was difficult because the materials and manipulations used in the treatment were totally different in the test and control groups. However, at least the outcome assessors should have been blinded. Because there was no blinding in these studies, there might be high risk of performance and detection biases.

Incomplete Outcome Data (Attrition Bias).
In the study by Claire et al. (2014) [32], the sample size was determined based on previous reported results. Intention-totreat analysis and dropout information were offered in this study also. erefore, there might be a low risk of attrition bias.
Information  [43]. Zhao et al. [43] reported one participant dropout into both groups. Liao et al. [41] reported two dropouts in the control group; however, the reason why those two patients were dropped out was not noted. Rao et al. [34] reported that three in the acupuncture therapy group (fear of pain), one in the ear-acupressure group (refused to continue treatment because of pain in the pinna), and four in the control group (severe headache, drowsiness and other adverse reactions occurred during taking the medicine could not continue to receive treatment) were dropped out. erefore, those three studies were at a low risk for incomplete outcome reporting bias. However, a crucial limitation was that intention-to-treat and per-protocol analyses were not conducted in these three trials.  [35][36][37][38][39][40]42] in terms of incompleteness bias because there were no sample size calculations, and no cases were reported to have been lost to follow-up or withdrawn from the trials. e incompleteness bias might be unclear in these trials due to failure to report dropouts.

Selective Reporting (Reporting Bias).
To reduce reporting bias, all analyses with and without statistically significant differences should be reported. One of the ways to assess reporting bias is to compare the results in the final reports with those in the protocol. However, 10 in 11 studies protocol could not be found in these studies, and only Claire et al. (2014) [32] declared a clinical trial registration number (ACTRN12608000149369). It was difficult to determine whether the other 10 outcomes were included in the published reports. Hence, the risk of reporting bias in these 10 studies was classified as "unclear," while the risk of reporting bias in Claire et al. (2014) [32] was low.

Other Potential Sources of Bias.
We set "support from pharmaceutical manufacturers" as other potential source of bias. Support in the form of free medical supplies, research Evidence-Based Complementary and Alternative Medicine         Evidence-Based Complementary and Alternative Medicine funding support, and medical supplies marketing cooperation coming from pharmaceutical manufacturers could bias results. Nevertheless, no other potential sources of bias were found in these studies. ere was publication bias by Rao et al. (2005). e risk of bias graph and summary of authors' judgements concerning included studies are shown in Figures 3 and 4, respectively.

Clinical Effectiveness.
According to raw data extracted from 11 original texts, primary outcomes including effective rate, total nasal symptom score, runny nasal score, sneezing score, global QoL score, and eye symptoms were presented. Adverse events were presented as secondary outcomes.

Total Effective Rate of EAP for AR.
e effective rate is a composite endpoint composed of improvement of clinical symptoms. e results can be divided into three categories: significantly effective, effective, and ineffective. Four versions of the Chinese AR Clinical Handbook Indicator recommend AR therapies [46][47][48][49]. ey were slightly different. In Haikou, 1997 criteria [46], "according to the sum of symptoms and signs scores before and after treatment, the efficacy of perennial allergic rhinitis was evaluated by the following formula: ≥51% was considered effective, 50%-21% was considered effective, and ≤20% was considered ineffective." In other reports, "according to the symptoms and signs score, the efficacy was evaluated by the following scoring methods: ≥66% was marked as effective, 65%-26% as effective, and ≤25% as ineffective." In our review, there are 10 trials reporting effectiveness rates [34][35][36][37][38][39][40][41][42][43] We pooled these trials using RevMan 5.3. A total of 869 participants were pooled, 433 in the treatment group and 436 in the control group. We used a random-effects model because of the significant heterogeneity (I 2 � 56%, P � 0.01). Meta-analysis revealed that the total effective rate in the treatment group was greater than that in the control group (pooled risk ratio � 0.51, 95% CI (0.36-0.70), P < 0.0001; Figure 5). Sensitivity analysis indicated that their result was robust ( Figure 6). Subgroup analysis showed that EAP was superior to control group treatments (RR: 0.63; 95% CI: 0.50-0.79; P � 0.005; Figure 7). e funnel plot (Figure 8) suggested that publication bias may exist; however, other factors could also be present in Rao et al. (2005). is may due to poor design, in particular, the poorly allocated concealment method (the allocation concealment method was not mentioned in this study), leading to exaggerated treatment effects.

EAP Related Adverse Events (AEs).
In these 12 studies, five did not mention adverse events information [35,36,[38][39][40], and six trials [32,34,37,[41][42][43] recorded adverse events. Of these, three studies [37,41,45] reported adverse events in two groups. Li et al. reported that five participants in the control group had nasal mucosa drying and bloody nasal mucus [42]. In the study by Rao et al. [33], the incidence of adverse reactions was reported as follows: no adverse reactions occurred in the acupuncture group (2.17% ear pressure group) or the control group (13.04%). Statistical analysis showed that there was a significant difference between the control group and the ear pressure group (P � 0.05). In the acupuncture group, a patient had dizziness and nausea during the process of acupuncture. After the needle was released quickly, it was completely relieved after resting for half an hour. ere were six cases of mild or transient adverse reactions, including two cases of mild headache and drowsiness, four cases of dry mouth, and three cases of gastrointestinal discomfort. However, how these effects were resolved was not noted.
In the study by Claire et al. [32], details of safety were noted specifically. ere were eight participants in the real group who reported 17 AEs and nine participants in the sham group who reported 20 AEs (x 2 � 0.01; P � 0.76). Some EAP-related AEs such as pellets irritating skin (one and two in real/sham groups) and ear acupoint inflammation (two and one events in real/sham groups) were reported during the 1 st week. ese events were effectively managed by refining the pressing techniques by the participants, without any medical assistance required. On the other hand, they reported that some participants reported headache or dizziness (11 and 14 events in the real/sham group) and insomnia (two events in the sham group).
Another study presented safety information; they reported that, after microwave therapy under nasal endoscope [32], there were seven complications in the control group (11.7%), including three cases of haemorrhage, two cases of nasal stenosis, and two cases of infection. Complications occurred in five cases (8.3%) in the group, including two cases of bleeding and infection each, and one case of nasal cavity stenosis. ere was no significant difference in the incidence of complications between the two groups. ey also noted that the treatment group had no obvious adverse reactions after auricular point pressing.

Overview of Findings.
To the best of our knowledge, there was only one previous systematic review published in 2010 on this topic [32]. Five studies were included in the previous review. ree of those were excluded by us because diagnostic criteria were absent in the original texts (Tables 3 and 4). By detailed comparison, we discovered that Rao and Han (2006) [50] in their review was Rao et al. (2005) [34] in our review. ese two studies illustrate the same experiment, with the same data and same author, however, in different publication years. In other words, this was a repeated publication. We chose Rao et al. (2005) instead of Rao and Han et al. (2006) because of the more detailed test records including laboratory instrument details, dropout details, and adverse reactions details. Hence, we have two studies identical to those of the previous review [34,36]. In the present review, we tried to update the topic based on the findings of the previous review.
We included 12 studies. e control group intervention can be classified in six categories: sham ear acupressure, conventional medicine therapy, acupuncture, electroacupuncture, Chinese medicine formula, and microwave therapy under nasal endoscope. EAP was not inferior to control group interventions (conventional medicine therapy, acupuncture, electroacupuncture, Chinese medicine formula, and microwave therapy under nasal endoscope) in terms of improving effective rate of allergic patients , symptom. However, the data extracted from 11 Chinese trials had small sample sizes and poor quality measures, according to the methodology measurement. e real EAP group was significantly greater than the sham group in terms of changes of global QoL score, scores for total nasal symptom, runny nose, and eye symptoms.
Validated questionnaire and scales such as the Quality Of Life Questionnaire are tools used to evaluate the quality of life of AR patients. e visual analogue scale is used to assess the severity of symptoms of AR. However, each has an application in included studies. Others such as quality of life score and nasal symptom scores evaluation methods can measure melioration of AR severity or disability; however, they are not widely used, despite the fact that these scales are recommended by the 2015 Clinical Guidelines.

Potential Biases in the Review Process.
No ongoing trials were found. e conclusion of this review was drawn from the 12 trials, comprising a limited number of participants.

Evidence-Based Complementary and Alternative Medicine
More studies and high-quality trials should be included in future reviews. In addition, 3 key points that may cause potential heterogeneity may be summarized as follows: As a noninvasive alternative, small seeds (Wang Bu Liu Xing and Vaccariae Semen) come from a plant or metal pellets on auricular points. Both are commonly used materials in EAP treatment and are effective. However, the differences between them remain unknown. In our review, Vaccariae Semen seeds applied in eleven China mainland publication trials [34][35][36][37][38][39][40][41][42][43]45] and stainless steel pellets (1.2 mm in diameter; PELSST S/Steel Tan; Acuneeds Co., Camberwell, Victoria, Australia) were used in a two centres (Royal Melbourne Institute of Technology University (Melbourne, Australia) and Clinical Trial Clinic and Guangdong Provincial Hospital of Chinese Medicine, Guangzhou, China) (Claire et al.) [32]. is situation might make a subtle difference in terms of efficacy. More in-depth studies on the two raw materials may be needed. Commonly used auricular points were summarized in our review: Nei Bi (TG4), lung (CO14), kidney (CO10), spleen (CO13), adrenal gland (TG2p), external nose (TG1, 2i), wind stream (SF1, 2i), and Shenmen (TF4). erefore, these can be regarded as commonly used EAR for AR ear points. However, in Lu et al. (2015) [39], three ear acupoints including internal nose (TG4), sympathetic (AH6a), and root of ear tragus (R2) were selected. is is very different from other studies, which may be the source of heterogeneity, because EAP is based on the meridians theory of TCM. In meridians theory, each acupoint serves a different purpose. e specificity of acupoints in morphological structure, biophysical characters, pathological reactions, acupuncture stimulation-induced responses in different brain regions, and therapeutic effects were supported by scholars [44].
Despite our use of validated effectiveness assessment criteria documents supporting trials in this review, the nonuniform standard of efficacy evaluation might influence outcomes and results (especially effectiveness rate). It might be challenging to employ the same diagnosis and effectiveness assessment criteria for each trial, as these criteria vary with each update.  Han (2006) Li (2018) Liao (2016) Lu (2015) Ouyang (2012) Rao (2005) Rao (2005a) Ye (2008) Yuan (2013) Yuan (2016) Zhao (2019) Lower CI limit Estimate Upper CI limit Meta-analysis estimates, given named study is omitted   Zhao (2019) Meta-analysis estimates, given named study is omitted Lower CI limit Estimate Upper CI limit Evidence-Based Complementary and Alternative Medicine

Conclusion
Despite the positive results of some outcomes, it is premature to confirm the efficacy of EAP for treating AR.
More high-quality studies are needed to validate its efficacy. ere are insufficient data to state that EAP is safe and reliable due to the small number of trials reporting adverse events. erefore, studies with larger sample sizes and rigorously designed studies are necessary to determine conclusively a definitive association between EAP and AR.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.  Figure 11: Forest plot for the runny nose score.