The Traditional Chinese Medicine and Relevant Treatment for the Efficacy and Safety of Atopic Dermatitis: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

2.1. Search Protocol

Because of the pervasive low literature quality in Chinese electronic databases, so the international mainstreamed electronic databases including PubMed, Embase, Science Direct, Medline, Web of Science, ProQuest, Springerlink, and Wiley Library Online were selected for articles extraction from their inception up to December 2016 by two reviewers’ (Zhao-feng Shi, Yi-quan Yan) cooperation. There were no language restrictions on literature searching in the selected database. The searching terms were conducted as follows: “allergic dermatitis”, “contact dermatitis”, “urticaria”, “drug eruption”, “eczema”, “traditional Chinese medicine”, “Chinese herbal medicine”, “Chinese medicine”, “acupuncture”, “moxibustion”, “random”, “randomized controlled trial”, and the related synonyms; meanwhile the term searching process also conducted the free text strategy and Medical Subject Headings (MeSH) term. The searching language was slightly adjusted based on the demand of different databases for adaptation. The hand searching process was also performed in the library of Fourth Military Medical University through the magazines to identify other potential related articles. Figure 1 clearly illustrated the literature identification, inclusion, and exclusion.

2.2. Studies Selection

This studies selection was conducted by two reviewers (Zhao-feng Shi, Tie-bing Song) to ensure the quality of searching. The disagreement was resolved by the third reviewers (Juan Xie) or repeated discussion. Eligible literatures were included in the meta-analysis if they suited the following criteria: (1) full-text randomized controlled trials; (2) participants of any age or gender diagnosed with AD or its relevant diseases (contact dermatitis, urticaria, drug eruption, eczema, etc.); (3) no active skin exudation or infection was found before experiment; (4) experimental group: patients who received traditional herbal medicine and related preparations or formula, acupuncture and moxibustion, and any treatment having the TCM characteristics, combined or not with the western medical therapy; (5) control group: participants who received placebo or western medicines; (6) all included patients having the need to stop the previous applied topical steroid or immunosuppressant at least 1 week before RCT was conducted; (7) more than 1 severe degree criterion of AD as follows: SCORAD, EASI, VAS, simple scoring system, the Children’s Dermatology Life Quality Index (CDLQI) [23], and itch intensity; (8) the major clinical outcome effectiveness of AD that was evaluated having one indicator at least as follows: (a) clinical effects; (b) the median clinical scores variation; (c) laboratory index; (d) the variation of symptom scores; (e) adverse effects; (f) lesion size change. Studies would be excluded if they do not meet the criteria mentioned above or the following: (1) severe clinical illnesses (such as liver or kidney disease) or infections; (2) absence or inconsistency of methods and results; (3) case report, comments, reviews, editorials, letters, and so forth.

2.3. Data Extraction and Quality Analysis

The literature data were extracted by two independent reviewers (Zhao-feng Shi, Yi-quan Yan) who deeply assessed the article inclusion and exclusion criteria based on the modern standardized procession. All discrepancies divided by two reviewers were judged by a third reviewer (Juan Xie) until they drew the final conclusion. The literature characteristics comprised the following items: (1) the name of author and the year of publication; (2) sample size (experimental group/control group); (3) participants’ age; (4) participants’ gender (experimental group/control group); (5) study design; (6) disease category; (7) diagnostic and evaluating criteria; (8) treating method; (9) treating duration; (10) dropout numbers; (11) major outcome parameter; (12) side effect. The authors of included articles were contacted if we could not obtain the literature characteristics mentioned before (Table 1).

Two independent investigators (Zhao-feng Shi, Tie-bing Song) conducted the study quality analysis based on the Cochrane Collaboration’s tools for the risk of bias assessment [24]. This tool contained six domains for the literature’s evaluation: (a) the random allocation method; (b) the allocation concealment; (c) the blinding method; (d) the outcome data integrity; (e) the outcome data of selective reports; (f) other bias sources. The study sheet was conducted which contained the evaluating criteria named “Y,” “N,” and “U” (“Y” refers to “yes,” “N” refers to “no,” and “U” refers to “unclear”). If one included study had three or more “Y,” this study could be recognized as being of high quality which has low risk of bias, and vice versa. The literature data were collected and analyzed efficiently by the software Review Manager (RevMan; version 5.2; the Nordic Cochrane Center, the Cochrane Collaboration, 2012 Copenhagen, Denmark) and presented at table (Table 2).

2.4. Statistical Analysis

Data of articles were pooled and analyzed by the software Review Manager (RevMan; version 5.2; the Nordic Cochrane Center, the Cochrane Collaboration, 2012 Copenhagen, Denmark). The risk ratio (RR) with 95% confidence interval (CI) for dichotomous outcomes and the Standard Mean Difference (Std. MD) or Mean Difference (MD) with 95% confidence interval (CI) for continuous outcomes was calculated, respectively, by the author (Zhao-feng Shi) and repeating verified by the author (Yi-quan Yan). The calculation of I² statistic was applied for the statistical heterogeneity for the purpose of explanation of potential inconsistency across the included studies. It was a quantitative tool that could provide an estimate of variation resulting from heterogeneity. If the result of I² statistic is between 25% and 50%, it would be recognized as being of low heterogeneity; the result between 50% and 75% would be of moderate heterogeneity and above 75% would be of high heterogeneity. The result of I² statistic that exceeded 50% was regarded as obtaining the heterogeneity according to the Cochrane Handbook (version 5.1.0) notation [25]. When the heterogeneity did not exist or was moderate, a fixed-effects model applying the Mantel-Haenszel method was performed to analyze data; when the heterogeneity was high, a random effects model applying the Der Simonian-Laired method was conducted [26, 27]. The subgroup and sensitive analysis was adopted on the condition of high data heterogeneity in order to find the source of heterogeneity. The publication bias was evaluated by funnel plot, Egger’s test, and Begg’s test through the Stata (version 11.0, StataCorp LP, College Station, US): the asymmetry of image in visual appearance or P value lower than 0.05 in Egger’s test or Begg’s test calculation could be considered as having the publication bias [28, 29].

3.1. The Characteristics and Quality Evaluation of Included Articles

Table 1 clearly illustrates details and characteristics of included 24 articles. From the table, articles were grouped based on the sample size, age, gender, study design, disease category, diagnosis and evaluating criteria, treating methods, treating duration, dropout, major outcome parameters, and side effects. These articles were all strictly randomized controlled trials (RCTs) and conducted in different areas around the world. Of the final included articles, one was performed in the United Kingdom (UK) [30], one in India [36], one in Japan [40], one in Korea [46], one in Iran [52], one in the United States [47], three in Germany [41, 44, 48], and fifteen in China [31–35, 37–39, 42, 43, 45, 49–51, 53]. The year of publishing for articles was between 1992 and 2016. Overall, five hundred and seventy-eight people with AD (experimental) and 533 people without AD (control) contributed to the meta-analysis, whereas 5 articles did not show the group dividing situation (507 people) [30, 31, 33, 41, 48, 51]. The age of included people disturbed by AD had a wide range from infant to elderly people. The diagnostic and evaluating criterion of AD in different centers did not reach an agreement and kept diversified. The SCORAD [34, 43, 44, 46, 48, 50, 51] and the Hanifin and Rajka criteria [36, 42, 43, 46, 51–53] were major diagnostic and evaluating criteria of studies. Traditional herbal medicine and related preparations [30, 31, 33–37, 40, 42, 43, 46, 49, 51, 53], acupuncture and related needling methods [32, 39, 41, 44, 45, 47, 48, 50], and Chinese herbal medicine extraction [36, 52] were the applied traditional Chinese medicines and related treatments of included articles in experimental groups. Treating duration of included articles was diversified ranging from 4 days to 5 months and all articles reported the treating duration. As for side effects, thirteen articles were not reported for any reasons [32, 38–41, 44–50, 53]. Our reviewers of meta-analysis had tried to contact the authors by e-mail but did not get useful information.

For all included studies in Table 2, random collection methods were perfectly presented and the major method was performed by random number table. Thirteen articles provided the allocation concealment method [30, 31, 34, 36, 40–44, 46–48, 51, 52] and eleven articles showed the blinding method [30, 31, 34, 40, 41, 43, 44, 46, 48, 51, 52]. All outcome data of included trials were integrity. Only one included article did not conform to the outcome data of selective report because the outcomes were merely presented by P value without specific outcome data supporting [39]. For the part of no other bias source, one article did not accord with it because there were a small number of subjects in each group, leaving the possibility that the results may have been due to chance or other bias [47]. Figures 2 and 3 were conducted by the software RevMan, both supporting the data of Table 2. Two reviewers (Zhao-feng Shi and Yi-quan Yan) conducted Tables 1 and 2 separately and disagreements of them were discussed with the third reviewer (Juan Xie) until the final same conclusion was made.

3.2. Meta-Analysis

3.2.1. Clinical Effectiveness

Figure 4 showed the clinical effectiveness of traditional Chinese medicines and relevant treatments for atopic dermatitis (AD) therapy. Eight articles which were all carried out in China including 667 participants were analyzed in the forest plot of meta-analysis. For the eight included articles, five of them (62.5%) [33, 35, 37, 38, 49] performed traditional Chinese herbal medicine in the experimental group while three of them (37.5%) [32, 45, 50] performed acupuncture related therapies (acupoint injection with autoblood [32], flying needle [50], and auricular acupuncture [45]). Five studies were published before 2010 and three were published after 2010. The heterogeneity of clinical effectiveness was moderately high (I² = 65%), so the random effects model was performed to calculate the combined data by Mantel-Haenszel test. The subgroup analysis was conducted based on the different treating methods (traditional Chinese medicine and related preparation, acupuncture, and related therapy) and publishing year of articles (before 2010, after 2010) to analyze the source of heterogeneity. Unfortunately, we still did not find the source of heterogeneity. The meta-analysis illustrated that the clinical effectiveness of AD for the experimental group which was performed by traditional Chinese medicines and relevant treatments was irrelevant to the control group (RR = 1.10, 95% CI = 0.99 to 1.21, and P = 0.07 > 0.05). There was no statistical difference for the result of clinical effectiveness.

3.2.2. SCORAD Amelioration

Figure 5 was the forest plot which illustrated the SCORing Atopic Dermatitis (SCORAD) amelioration after traditional Chinese medicines and relevant treatments for experimental groups and western medicine or placebo treatments for control groups. Four articles including 173 patients were collected and analyzed in the forest plot of meta-analysis [43, 44, 46, 50]. Standard Mean Difference (Std. MD) was conducted as a combined statistics for the cause of the obvious difference of standard deviation (SD) in the same articles’ SCORAD result. Of the included studies, two (50%) conducted traditional Chinese herbal medicine as experimental method [43, 46] and others (50%) conducted acupuncture [44] and flying needle [50]. The heterogeneity was so high (I² = 86%) that the random effects model was conducted to calculate the combined data by inverse variance (IV) approach. The subgroup analysis was performed based on the treating methods (traditional Chinese medicine and related preparation, acupuncture and related therapy) to find the source of heterogeneity. However, the result still could not provide the certain conclusion. The meta-analysis showed that the SCORAD amelioration for the experimental group and the control group was irrelevant (Std. MD = 0.89, 95% CI = −0.24 to 2.02, and P = 0.12 > 0.05). The result of SCORAD amelioration had no statistical difference.

3.2.3. EASI Amelioration

Figure 6 showed the Eczema Area and Severity Index (EASI) amelioration after traditional Chinese medicines and relevant treatments for experimental groups and western medicine or placebo treatments for control groups. Only 2 studies which incorporated 50 participants were analyzed in meta-analysis [46, 53] and Mean Difference (MD) was conducted as a combined statistics. Both the included trials chose traditional Chinese herbal medicine in the experimental group for AD therapy and were published in recent years. There was no heterogeneity (I² = 0%) in the meta-analysis, so the fixed-effects model was conducted to calculate the combined data by inverse variance (IV) approach. The meta-analysis revealed that traditional Chinese herbal medicine was superior to western medicine for EASI amelioration in AD (Std. MD = 3.22, 95% CI = 0.41 to 6.03, and P = 0.02 < 0.05). The result had statistical difference.

3.2.4. SSRI Amelioration

Figure 7 illustrated the Symptom Score Reducing Index (SSRI) amelioration after traditional Chinese medicines or related treatments for experimental groups and western medicine or placebo treatments for control groups. Only 2 appropriate articles including 105 patients were collected and analyzed in meta-analysis [35, 38]. Standard Mean Difference (Std. MD) was conducted as a combined statistics for the cause of the obvious difference of standard deviation (SD) of the same article in SSRI result. Two included articles both used traditional Chinese herbal medicines in the experimental group for AD treatment. We found that the heterogeneity of included studies was obvious (I² = 76%) maybe because of the variation of Chinese medicine ingredients between two trials and measurement time. The random effects model was conducted to calculate the combined data by inverse variance (IV) approach. The meta-analysis showed that the SSRI amelioration for the experimental group and the control group was irrelevant (Std. MD = −0.36, 95% CI = −1.16 to 0.45, and P = 0.39 > 0.05). The result of SSRI amelioration had no statistical difference.

3.2.5. Funnel Plot Characteristics

The funnel plot (Figure 8) was drawn by reviewer (Zhao-feng Shi) based on pooled odds ratio (OR) as the midpoint. The publication bias of clinical effectiveness in included studies was evaluated by comparing the symmetry of the funnel plot. The symmetry of funnel plot was assessed by two reviewers’ cooperation (Juan Xie and Yi-quan Yan) in the visual aspect and both the reviewers considered that this image was symmetrical, which means the clinical effectiveness did not have the publication bias. We also conducted Egger’s test and Begg’s test to confirm whether the publication did exist. The calculating result of Egger’s (t = 1.36, P = 0.247) and Begg’s test (z = 1.88, P = 0.06) and images indicated that the publication bias did not exist [54] (Figure 9).

4.1. Summary of the Main Evidence

This systematic review and meta-analysis have shown the clinical effectiveness of traditional Chinese medicine and related treatment and the amelioration of related evaluating criteria for atopic dermatitis (AD) therapy by collecting 24 high quality articles from 8 main international electronic databases. The evidence-based rules of traditional Chinese medicine and related treatment for AD have been conducted by this systematic review and meta-analysis. This study has four obvious advantages: (a) articles having good quality based on the Cochrane Collaboration’s tools for the risk of bias assessment were included ensuring that the results trustworthy; (b) the articles’ quality was evaluated by three reviewers’ cooperation (Zhao-feng Shi, Juan Xie and Yi-quan Yan); (c) the subgroup analysis was well performed to find the source of heterogeneity; (d) the author was contacted by our reviewers if an appropriate study cannot be acquired as full-text; (e) the funnel plot, Egger’s test, and Begg’s test were conducted to evaluate the publication bias of articles’ clinical effectiveness.

The result of the clinical effectiveness of traditional Chinese medicines and relevant treatments for atopic dermatitis (AD) therapy in meta-analysis showed that the experimental group (traditional Chinese medicine and relevant treatment) was irrelevant to the control group (placebo or western medicine). The subgroup analysis did not reveal the source of heterogeneity. The clinical evaluating criteria amelioration of SCORAD and SSRI for the traditional Chinese medicines and relevant treatments of AD provided the same conclusion as the clinical effectiveness. However, the result of EASI in meta-analysis illustrated that the experimental group (traditional Chinese medicine and relevant treatment) was superior to the control group (placebo or western medicine). Egger’s and Begg’s test and funnel plot did not reveal the publication bias for clinical effectiveness. In a word, under this circumstance, we could not give the firm conclusion on the effectiveness of traditional Chinese medicines and relevant treatments for AD except that we can include more relevant articles and credible results in the future meta-analysis.

4.2. The AD Adjunctive Therapy for Recommendations

Atopic dermatitis (AD) was recognized as a common inflammatory skin disease nowadays that bring much burden to humankind around the world, with the reported rate as high as 30% of children and 10% of adults [1]. Despite the obvious adverse effects (deranged metabolism, growth suppression, increased susceptibility to infection, and suppression of the hypothalamic-pituitary-adrenal axis) [55, 56], western medicines that included skin emollients, corticosteroids, antihistamines, and immunomodulation agents still have been put forward as AD standard therapy for a long time. Under the characteristics of a chronic course and repeated attack of AD, clinical physicians tried their best to find appropriate adjunctive treatments for better clinical control of AD symptoms. Traditional Chinese medicines and related treatments are popular examples of AD adjunctive therapy, which could be grouped in Complementary and Alternative Medical (CAM) treatments category.

In many Asian countries, such as China, Japan, South Korea, India, and Iran, traditional Chinese medicine (TCM) had been applied on clinical areas of AD for its outstanding efficacy and safety. TCM may have the ability to downregulate cells that associated with AD through antioxidant activities [57]. In addition to that, TCM may affect the structure of the developing dendritic cells (DCs) through losing their typical dendritic morphology and decreasing their expression of CD1a as well as the low affinity IgE receptor CD23 [58]. A China TCM Expert Consensus of AD was published in 2013, separating AD as four main syndromes based on theory of TCM: firstly, the brimming heat of heart and spleen syndrome, conducting the Chinese herbal prescription named Sanxindaochiyin; secondly, the heat of heart and spleen deficiency syndrome, conducting Qinxinpeitufang; thirdly, the spleen deficiency and dampness retention syndrome, performing Xiaoerhuashitang; fourthly, the syndrome of wind and dryness due to blood deficiency, conducting Dangguiyinzi [59]. However, there was still lack of modern randomized controlled trials to cautiously evaluate the TCM treatment of AD and the pharmacological mechanism of TCM for AD has been difficult to explain. Acupuncture and the relevant therapies (such as flying needle, auricular needle, and electroacupuncture) have been shown to be practical in patients with AD that may be due to the antipruritic effect of kappa-opioid receptor activated by high-frequency stimulation [60]. However, since many of them lack scientific verification, they should be performed under the supervision of physicians [12].

4.3. Limitation

This systematic review and meta-analysis had several potential limitations restricting the clinical application for AD: (a) Included studies are smaller than we expected although reviewers restricted the electric database numbers for high quality articles. (b) The number of participants in some included studies is small: Pfab. F only included 10 patients in their studies and Lee K.C.’s article only contains 15 patients [44, 47]. We consider that the limited participant numbers may be insufficient to evaluate the efficiency and safety of AD therapy. (c) The heterogeneity of included studies is obvious although reviewers conducted strictly study selection, data extraction, and quality analysis. And the subgroup analysis of studies also could not clearly reveal the source of heterogeneity. We estimate that the heterogeneity may come from the sample size, medicine application and dose, and publication years, as well as treating duration. We are required to understand that the random effects model was performed to pool data which cannot give exact and stable conclusion in this situation. (d) The traditional Chinese medicines and related treatments are diversified in the included articles: some studies used traditional Chinese herbal medicines and some studies performed acupuncture and related treatments. Also, therapy in control group could not reach a consensus as a uniform method.