Efficacy and Safety of Banxia Formulae for Insomnia: A Systematic Review and Meta-Analysis of High-Quality Randomized Controlled Trials

Objective To systematically evaluate the efficacy and safety of Banxia (Pinellia Tuber) formulae in the treatment of insomnia compared with those of conventional western medicines. Methods Randomized controlled trials (RCTs) evaluating the efficacy and safety of Banxia formulae in the treatment of insomnia were searched from the following databases: PubMed, Cochrane Library, EMBASE, the China National Knowledge Infrastructure (CNKI), Chinese Scientific Journals Database (VIP), and Wanfang database. The literature collected was from the time when the databases were established to April 2020. Quality assessment and meta-analysis were conducted by using Cochrane bias risk assessment tool and RevMan 5.2, respectively. Publication bias was assessed by Egger's test. Results Fourteen RCTs with 910 participants were identified. A total of 46 traditional Chinese medicines involving 2 different dosage forms were used in the included studies. Meta-analysis indicated that Banxia formulae had more significant effects on improving the total effective rate (RR = 1.23, 95% CI 1.16 to 1.31), Pittsburgh Sleep Quality Index (PSQI, MD = −1.05, 95% CI −1.63 to −0.47), and the TCM syndrome score (SMD = −0.78, 95% CI −1.18 to −0.39). Meanwhile, on reducing adverse events, Banxia formulae also showed an advantage (RR = 0.48, 95% CI 0.24 to 0.93). Conclusion According to the current studies, the efficacy of Banxia formulae in the treatment of insomnia is better than that of the conventional western medicines, and its safety is relatively stable. However, due to the limitations of this study, further research and evaluation are needed.


Introduction
Insomnia is a widespread health complaint in general population which deserves appropriate recognition and attention. It is not only a risk factor for cardiovascular diseases (such as arterial hypertension, chronic heart failure, and myocardial infarction), obesity and type 2 diabetes, but also a common disease which frequently coexists with nervous system diseases (dementia, depression) [1,2].
Cognitive behavior therapy and drug therapy are the main methods of treating insomnia. Since cognitive behavior therapy does not work for everyone in all situations, drug therapy is still the most common method for treating insomnia so far [3,4]. In western medicine, short/intermediate acting benzodiazepines (BZ), benzodiazepine receptor agonists (BZRA), and some antidepressants are drugs of the first choice for the treatment of insomnia. Although western medicines have quick and strong effects in the treatment of insomnia, many side effects of them have been reported, including hangover, nocturnal confusion, falls, rebound insomnia, tolerance and dependency liability, withdrawal reaction, and increased risk of slight infection. [5][6][7]. Studies have shown that BZ and BZRA can impair one's driving ability which has significant correlation with car accidents [8][9][10]. Moreover, some data in current studies suggests that the mortality of people who use BZ, antidepressants, and antipsychotics is increased [11,12]. ese side effects have caused many medical and social problems.
Chinese medicine has a long history in the treatment of insomnia. Based on the conception of holism and the theory of syndrome differentiation and treatment, traditional Chinese medicine treats insomnia and regulates the patient's body as a whole. It has unique advantages in improving sleep and life quality of the patients, and has little side effect.
Banxia (Pinellia Tuber) is the dried tuber of Pinellia ternata ( unb.) Makino. As a Chinese herbal medicine, Banxia has been used in the treatment of insomnia for about two thousand years. As early as in "Huangdi's Internal Classic," there were records of Banxia for insomnia treatment. In materia medica works in past dynasties of China, there are also records of Banxia for insomnia treatment. For example, in "Compendium of Materia Medica," it said: "It (Banxia) can remove water distention and treat insomnia." According to modern pharmacological studies, Banxia has antitussive, expectorant, antitumor, antibacterial, anti-inflammatory, antioxidant and sedative-hypnotic effects [13]. It is commonly used in the treatment of cough, vomiting, infection, inflammation, and emotional illness.
Previous meta-analysis [14] and systematic review [15] of Huanglian Wendan Decoction in the treatment of insomnia were limited by considerable risk of bias and a comprehensive meta-analysis on the efficacy of Banxia formulae in the treatment of insomnia is lacking. erefore, the purpose of the present study was to systematically review randomized controlled trials (RCTs) of high-quality which investigate the efficacy and safety of Banxia formulae compared with those of conventional western medicines in insomnia adults.

Method
is systematic review and meta-analysis are based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses: PRISMA statement search strategy [16].

Search Strategy.
Six databases including PubMed, Cochrane library, EMBASE, the China National Knowledge Infrastructure (CNKI), Chinese Scientific Journals Database (VIP), and Wanfang databases were searched from inception to April 2020 for the relevant RCTs of Banxia formulae for insomnia, with "insomnia" and "traditional Chinese medicine" as search terms, a subject word plus free words as search form. In order to ensure that eligible herbal formulae were included as many as possible, the specific herb name "Banxia" was not explicitly searched, and no restriction on language and publication period was set in this review. In addition, we also searched for references that have been included in relevant literature or systematic review.

Inclusion Criteria
(1) Type of Participants. Adult patients diagnosed with insomnia were included. Insomnia was confirmed according to standard diagnostic criteria including the "Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-4)" [17] and the "Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5)" [18]. (2) Type of Study. Only RCTs that assessed the efficacy and safety of drugs for the treatment of insomnia were eligible. Trials that only mentioned "randomization" but without any description of the random allocation process were excluded. e titles and abstracts of the retrieved articles were read by 2 independent review authors to exclude the obvious disqualified RCTs. en, the full texts of the studies that potentially met the predefined inclusion criteria were obtained and read in order to select the eligible RCTs. When there were different opinions, the 2 review authors can reach consensus by discussing with the corresponding author of this article.

Quality Assessment.
e methodological quality of the included RCTs was assessed by 2 independent review authors according to the Cochrane RoB2 criteria [19]. e signalling questions used to evaluate bias in each trial related to five domains: (1) bias arising from the randomization process; (2) bias due to deviations from intended interventions; (3) bias due to missing outcome data; (4) bias in measurement of the outcome; (5) bias in selection of the reported result. By responding to signalling questions, each domain was judged as low risk of bias, some concerns, and high risk of bias. en, the overall bias of each trial was determined. e disagreements on the methodological quality between the 2 review authors were solved by discussion and consulting a third author.

Data Extraction.
e data extraction was performed independently by 2 review authors according to the predesigned standard data extraction forms including the following items: lead author, publication year, country of origin, characteristics of participant, course of treatment, adverse events, and outcome measures. Disagreements were resolved in line with the principle of consensus through consultation. If the study reported the outcome data at different time points, the data of the last time point was extracted. For the trials with more than 2 groups or factorial designs allowing multiple comparisons, only the information and data reported in the original articles were extracted.

Banxia Formulae Composition.
In each included study, the Banxia formula and its composition were recorded, with a frequency analysis of common drugs combined with Banxia.

Data Analysis.
e mean difference (MD) or standardized mean difference (SMD) was used to evaluate continuous data, and relative risk (RR) with 95% confidence intervals (CI) for dichotomous data. In order to evaluate whether the efficacy of Banxia formulae was affected by variable factors, subgroup analyses were conducted according to the course of treatment (≤21 d, >21 d), the dosage of Banxia (≤9 g, >9 g), and the processing method of Banxia (raw Banxia, processed Banxia). Based on the analysis, whether the differences among the subgroups were statistically significant was assessed. All meta-analyses in this study were performed by using the software Cochrane Collaboration Review Manage (RevMan 5.2), and P < 0.05 was considered statistically significant. Egger's test was carried out by Stata 15.1.

Heterogeneity.
e statistical heterogeneity of the trials was assessed by X 2 test and expressed as I 2 value. When there is no heterogeneity or the heterogeneity was moderate (P > 0.1, I 2 < 50%), a fixed effect model (FEM) was applied; otherwise, a random effect model (REM) was applied.

Publication Bias.
e publication bias was identified by Egger's test, if an outcome was reported in at least 10 trials.

Sensitivity Analysis.
e sensitivity analysis was performed by excluding one study from the meta-analysis. If the estimated value of the point exceeded the 95% CI of the total effect amount (or was significantly different from the combined effect amount) after the study was excluded, it indicated that there were potential risks and this study needs to be further reviewed.

Description of Studies.
A total of 2523 studies were retrieved from the 6 electronic databases. After the duplicates were removed, 1956 studies remained. In the remaining 1956 studies, 122 were eligible. By reading the full text of the 122 studies, 108 were excluded, among which 24 studies were with inappropriate interventions, 43 studies were without control group, 1 study was without full text, 7 studies were not real RCTs, 25 studies did not use Banxia formula, 7 studies had high risk of bias according to Cochrane RoB2, and 1 study was repeated publication. Finally, 14 studies including 910 participants were included and a meta-analysis was conducted on them [20][21][22][23][24][25][26][27][28][29][30][31][32][33]. A PRISMA flowed diagram shows the procedure of literature research and study selection ( Figure 1).

Basic Characteristics of the Included Studies.
Basic characteristics of the 14 included studies are summarized in Table 1. All the eligible studies were conducted and published in China. All the included studies evaluated the total effective rate, among which 11 RCTs showed that Banxia formula improved the total effective rate [20-22, 25-27, 29-33], 3 RCTs showed that Banxia formulae did not significantly improve the total effective rate compared with conventional western medicine [23,24,28]. In the 6 RCTs with PSQI scores, 4 showed that Banxia formula improved the total score of PSQI of patients with insomnia [22][23][24]32], and the other 2 showed that Banxia formula did not significantly improve the total score of PSQI of patients with insomnia [26,28]. In the included studies, 5 studies reported TCM syndrome score [22-24, 28, 32], and 8 studies described side effects [20, 22-26, 29, 32]. e treatment duration ranged from 7 days to 30 days (Table 1).

Description of Banxia Formulae.
e composition of Banxia formulae in the included studies is listed in Table 2. is paper mainly introduced the basic composition of Banxia formulae in the included studies. A total of 46 herbs were used in the 14 included Banxia formulae, and 2 dosage forms were mentioned, including decoction (n � 13) and decoct-free granule (n � 1) [30] (Table 2). Among the decocting methods of the 13 decoctions, one was decocting Banxia for a long time [33] and the others were conventional decocting. e top 7 frequently used Chinese herbal medicines were Banxia (Pinellia ternata ( unb.) Makino, frequency � 14), Gancao (Glycyrrhiza uralensis Fisch. ex DC., frequency � 10), Chenpi (Citrus × aurantium L., frequency � 9), Fuling (Poria cocos (Schw.) Wolf, frequency � 98), Zhuru (Bambusa beecheyana Munro, frequency � 8), Huanglian (Coptis chinensis Franch., frequency � 7), and Dangshen (Codonopsis pilosula (Franch.) Nannf., frequency � 6) ( Table 3). Figure 2 shows the assessment of the risk of bias. All the included studies were described as "randomized." e random sequence was generated by random number table method (11 studies) or computer-generated random number Evidence-Based Complementary and Alternative Medicine 3 method (3 studies). One study applied "sealed envelopes," two studies mentioned single-blind method, and the remaining 11 studies had no clear information of blind method. ere were no imbalances in the 14 studies. erefore, there was no high risk in the randomization process. Because of vague description of blind method, 11 trials were judged as some concerns. Two studies described dropouts and provided adequate explanations, while the other 12 studies did not mention dropouts. e method of outcome measurement was suitable for the outcome they were intended to evaluate in each trial. No other significant bias was found in the included studies.

Efficacy Assessment
3.5.1. Total Effective Rate. All the 14 included studies reported the total effective rate, which indicated the total effective rate of the experimental group was higher than that of the control group (RR � 1.23,95% CI 1.16 to 1.31) (Figure 3). In the heterogeneity test, P � 0.14, I 2 � 29%, so the FEM was used to conduct the statistical. Pooled RR with 95% CI showed Z � 6.40, P < 0.00001 (Figure 3), suggesting that the difference was statistically significant. It can be concluded that the total effective rate of the experimental group with Banxia formulae was higher than that of the control group with conventional western treatment.

Total Score of PSQI.
Six studies reported the total score of PSQI. e results varied with the course of treatment.       (Figure 7).   Figure 8).

Adverse Event(s).
Eight studies mentioned adverse events, in which 5 reported adverse events occurring during treatment and provided sufficient information about the adverse events [20,23,26,29,32]. A total of 6.5% (10/154) patients in the experimental group and 14.9% (22/148) patients in the control group suffered from adverse events ( Table 5). Some patients had multiple events. ree studies declared there were no serious adverse events or side effects in the experimental group and the control group [22,24,25].
Adverse events in the experimental groups mainly included numbness in the lips or the tongue, throat discomfort, bitter taste in the mouth, dry mouth, nausea, acid reflux, heartburn, diarrhea, etc. Adverse events in the control group mainly included daytime drowsiness, dizziness, fatigue, dry mouth, bitter taste in the mouth, constipation, upset, drowsiness, deterioration of memory, and loss of appetite. ere were no reports of any serious adverse events related to drugs used in the studies.

Publication Bias.
In the Egger's test for evaluating publication bias in total effective rate of the two groups, P � 0.012, which indicated there was certain publication bias ( Figure 10). For studies on the other outcomes,  Test for overall effect: Z = 6.40 (P < 0.00001) Heterogeneity: chi 2 = 18.43, df = 13 (P = 0.14); I 2 = 29% Evidence-Based Complementary and Alternative Medicine because the number was less than 10, Egger's test was not performed.

Sensitivity Analysis.
e sensitivity analysis did not indicate that the final outcome would change because of the exclusion of any individual study, which meant that no single study can significantly affect the pooled RR and 95% CI.

Summary of Evidence.
Banxia is a classic Chinese medicine with a long history of treating insomnia. Many clinical trials and pharmacological studies [34][35][36] as well as some reviews [37,38] have provided evidence for the efficacy of Banxia in treatment of insomnia. However, there is no meta-analysis on the value of Banxia formulae for insomnia. is paper is a systematic review of 14 high-quality RCTs including 910 participants aiming at determining the efficacy and safety of Banxia formulae in treatment of insomnia compared with conventional western medicines. rough the study, we found that Banxia formulae provided statistically significant benefits in improving the total effective rate (RR � 1.23, 95% CI 1.16 to 1.31), the PSQI (MD � −1.05, 95% CI −1.63 to −0.47) and the TCM syndrome score (SMD � −0.78, 95% CI −1.18 to −0.39), as well as in decreasing the incidence of adverse events (RR � 0.48, 95% CI 0.24 to 0.93).
e results of this meta-analysis showed Banxia formulae, compared with conventional western medicines, could significantly improve the total effective rate of insomnia patients, regardless of the course of treatment, the dosage, and proceeding method of Banxia. When the course of treatment is longer than 21 days, Banxia formulae can significantly improve the PSQI of insomnia patients. Although 5 studies including 154 patients reported adverse events, only 10 patients had mild adverse events which was possibly related to Banxia formulae without powerful evidence. e meta-analysis showed that there was statistically significant difference in the incidence of adverse events between the two groups (RR � 0.48, 95% CI 0.24 to 0.93, P � 0.03, heterogeneity χ2 � 2.68, P � 0.61, I 2 � 0%). erefore, current studies indicate the efficacy of Banxia formulae in treatment of insomnia is stable, without adverse reactions.

Limitations. Some limitations existed in this study.
Firstly, despite the efforts by us to search the commonly used databases at home and abroad as comprehensively as possible, all the finally included studies were conducted and published in China, because traditional Chinese medicine has not been widely promoted and used in other countries. erefore, the results of this study have certain limitations. Secondly, although trials with high risk of bias were excluded based on RoB2, some included studies were still short on methodological details. Only two studies described an  appropriate method of allocation concealment, and only two studies used single-blind method. Due to the condition limitations, the current designs for RCTs on Banxia formulae could not meet the demand of blind method compared with those for RCTs on conventional therapy. irdly, Egger's test (P � 0. 012) indicated the existence of publication bias, so the efficacy of Banxia formulae on insomnia may be overstated. erefore, in this metaanalysis, the efficacy and safety of Banxia formula in the treatment of insomnia are based on the current research data, and more high-quality, multi-center, large sample RCTs are needed for further evaluation.

Implications for Practice.
Modern pharmacological studies show that Banxia or Banxia preparations have sedative and hypnotic effects. ere are also some studies about the mechanism of Banxia. Wu et al. found that ethanol fraction from Rhizoma Pinelliae Praeparatum possessed sedative, hypnotic, and anticonvulsant activities and these activities may be related to the GABAergic system [36]. Lin et al. reported that Banxia Preparation (raw Banxia processed with liquor rice, lime and alum as adjuvant materials) reduced wakefulness and increased sleep in mice by increasing the number of random eye movement sleep episodes, the number of transitions from non-random eye movement sleep to random eye movement sleep and from random eye movement sleep to wakefulness [34]. Banxia has irritative toxicity to oral cavity, throat, gastrointestinal mucosa, and cardiotoxicity, which can cause vomiting, diarrhea, fetal deformity and death, inflammatory reaction, and liver damage [39][40][41]. In the Pharmacopoeia of People's Republic of China, the prescribed dosage of Banxia is 3-9g [42]. Studies have shown the toxic constituent of Banxia is special acicular crystal which is called raphides. e poisonous raphides are mainly composed of calcium oxalate, proteins, and microamount of polysaccharides [43]. e irritative toxicity of poisonous raphides is related to macrophages [44]. e mechanism is that poisonous raphides penetrate into tissues to activate resident macrophages, release phagocytic and inflammatory cytokines, and cause mass migration of neutrophil, which finally leads to strong acute inflammatory response [45]. Banxia combined with ginger or used in its preparation forms (such as Fabanxia, Jiangbanxia, and Qingbanxia) can reduce the toxicity of Banxia [40,46,47].
In the included studies, there were 5 trials with Banxia and 9 trials with Banxia preparation (6 Fabanxia, 2 Jiangbanxia, 1 Qingbanxia). As for the dosage of Banxia, 4 trials were ≤9 g and 10 trials were >9 g, which showed the clinical application habit of Banxia. In this study, we found Banxia formulae can significantly improve the total effective rate of patients with insomnia, regardless of the dosage and processing method of Banxia, and the safety of Banxia formulae for insomnia is relatively stable. However, due to the lack of description of adverse events in the outcome, we did not   Test for overall effect: Z = 4.11 (P < 0.0001) Heterogeneity: chi 2 = 0.00, df = 1 (P = 0.98); I 2 = 0%      Banxia should be used with ginger or used in processed forms.

Conclusion
According to the current studies, the efficacy of Banxia formula in the treatment of insomnia is better than that of the conventional western medicines, and its safety is relatively stable. Current evidence-supported Banxia formulae could be used as a beneficial treatment for insomnia. However, due to the limitations of this study, further research and evaluation are needed.

Disclosure
Yan-Hua Lin and Cong Chen are the co-first authors.

Conflicts of Interest
e authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Authors' Contributions
Yan-Hua Lin and Cong Chen contributed equally to this work. S-YM developed the research ideas and then designed the search strategy. L-YH, ZX, and X-GX did data acquisition. L-YH, CC, and Y-MQ analyzed data. L-YH wrote the first draft of the manuscript. M-YF contributed to the revision. All authors interpreted the data analysis and critically revised the manuscript.