Irritable bowel syndrome (IBS) is a disease with a high incidence rate, and diarrhoea-predominant irritable bowel syndrome (IBS-D) is a subtype of irritable bowel syndrome with a major clinical manifestation. IBS has a prevalence ranging from 1.1 to 29.2% in the whole population according to the Rome III criteria, with the diarrhoea-predominant type accounting for about 23.4% [
Now, more and more studies use sham acupuncture as the control of acupuncture. However, there is a debate on whether sham acupuncture has curative effect and to what extent sham acupuncture does affect the final result; this question could be solved with the Network Meta-Analysis.
In this study, by collecting previously published treatments of IBS-D in randomized controlled treatment studies using acupuncture and oral common drugs, we expected to determine the following issues: (1) a ranking of acupuncture and drugs in the treatment of diarrhoea-predominant irritable bowel syndrome; (2) a ranking of acupuncture and drugs in their side effects on diarrhoea-predominant irritable bowel syndrome; (3) the extent to which sham acupuncture does effect the final result; (4) the acupoint distributions used to treat diarrhoea-predominant irritable bowel syndrome.
We conducted a standardized report based on the preferred reporting items of the PRISMA statement [
We searched PubMed, the Cochrane Library, Embase, and 4 Chinese databases [China National Knowledge Infrastructure (CNKI), Wanfang Database, VIP Database, and Chinese Biomedical Database (CBM)] to conduct a comprehensive database retrieval using a (acupuncture or electro-acupuncture Or acupuncture, Sham Acupuncture, pinaverium bromide, alosetron, eluxadoline, ramosetron, rifaximin), (randomized controlled trials or randomized controlled trials or clinical trials), and (IBS-D) strategy (the retrieval time was from the building of database to
We included randomized controlled trials that met the following eligibility criteria: (i) adult patients; (ii) single drug use; (iii) clinical trials with treatment duration greater than two weeks; (iv) articles that were not comments or commentary; and (v) patients that did not suffer from pregnancy or lactation, peptic ulcer, rectal disease, or liver or other systemic disease and had no previous history of gastroduodenal surgery or brain disease or surgery.
Articles were independently screened by two researchers. Initially, NoteExpress software (Beijing Aegean Sea Music Technology Co., Ltd.) was used to delete duplicate records. The remaining summaries and full texts were reviewed on the basis of inclusion and exclusion criteria, and disagreements were resolved through discussion.
Two reviewers (Lingping Zhu and Shasha Ye) independently extracted the relevant information from each eligible study based on a pre-prepared data abstraction sheet. Data included the location and study design of the trials, clinical characteristics, number of patients, patient age, diagnostic methods, treatment duration, outcome data, and side effects. The quality of the included studies was assessed using the Jadad scale, including three items such as randomized (2 points), double-blinded (2 points), and withdrawals and drop-outs (1 point) [
The primary outcome was the number of people who showed effective treatment, with secondary outcomes including side effects and common acupuncture points. Common side effects included constipation and rash.
The assessments of acupuncture and drug efficacy were based on a combination of the data extracted from the included trials, and then direct and indirect comparisons were used to assess the overall effect of acupuncture and medications. In this meta-analysis of the network, we used a random-effects model in a Bayesian framework. The odds ratio (OR) and 95 % confidence interval (CI) were used to analyse the effects of acupuncture and drugs on the efficacy of diarrhoea-predominant irritable bowel syndrome. CIs with OR> 1.0 indicated high risk, and CIs not containing 1.0 were considered statistically significant. All analyses used the GeMTC package generated by R software [
Node-splitting models were used to assess the consistency of the meta-analysis of the network to test whether the results of the direct and indirect comparisons were consistent within the treatment cycle [
A mesh diagram, contribution graphs, and publication bias tests were drawn using STATA 14.0 software (Stata Corporation, College Station, TX, USA).
A total of 1119 articles were obtained from the system search. After reviewing the literature, 40 duplicates were deleted. In addition, due to discrepancies in inclusion criteria, 1046 articles were excluded. Finally, a total of 33 trials were identified (Figure
Characteristics of included studies. NS means no available data.
Publication Date | Author | Experiment group (n) | Control group (n) | Treatments versus Control | Age of experiment group | Age of control | Diagnosis | Diagnosis criteria | Experiment Events | Control Events | Treatment Duration | Jadad | Gender | Nation | Side Effect€ | Side Effect © | Assessment tool |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2015 | LI Xueqing | 30 | 30 | NS/pinaverium bromide 50mg tid | 46±16 | 44±16 | IBS-D | ROME III | 28 | 24 | 8 weeks | 2 | Mixed | China | 0 | 0 | symptom assessment tool (China) (4 points) |
2014 | Zhan Daowei | 29 | 28 | (LR3, ST36,SP6,ST25, ST37,GV20,EX-HN3) |
42±14 | 37±13 | IBS-D | ROME III | 26 | 19 | 4 weeks | 3 | Mixed | China | 0 | 0 | symptom assessment tool (China) (4 points) |
2014 | Kong Suping | 29 | 28 | (GV20,CV12, ST25, ST36, SP9, ST39) |
38±11 | 38±11 | IBS-D | ROME III | 26 | 23 | 4 weeks | 3 | Mixed | China | 0 | 0 | symptom assessment tool (China) (4 points) |
2014 | Liu Shuying | 30 | 30 | (GV20,EX-HN3,CV12,ST25,ST37,ST39) |
41.4±11.8 | 41.77±8.99 | IBS-D | ROME III | 27 | 23 | 4 weeks | 1 | Mixed | China | 0 | 0 | symptom assessment tool (China) (4 points) |
2013 | Wu Yuanjian | 30 | 30 | (ST25, ST36, ST37, SP6, CV8) |
37.9±10.2 | 39.8±11.2 | IBS-D | ROME III | 26 | 24 | 4 weeks | 1 | Mixed | China | 0 | 0 | symptom assessment tool (China) (4 points) |
2012 | Pei Lixia | 30 | 30 | (ST25, ST36, ST37, SP6, LR3, GV20, EX-HN3) |
40.9±10.6 | 37.93±11.45 | IBS-D | ROME III | 27 | 24 | 4 weeks | 3 | Mixed | China | 0 | 0 | symptom assessment tool (China) (4 points) |
2013 | LI HAO | 35 | 35 | (ST 25, ST 36, ST37, SP6, LR3, GV20, GV29) |
37.9±11.5 | 39.1±11.8 | IBS-D | ROME III | 33 | 27 | 4 weeks | 5 | Mixed | China | 0 | 0 | symptom assessment tool (China) (4 points) |
2011 | Sun | 30 | 30 | (ST 25, ST 36, SP6,LR3,DU20,EX-HN 3 and ST 37) |
38.81±11.8 | 38.59±11.45 | IBS-D | ROME III | 27 | 24 | 4 weeks | 3 | Mixed | China | 0 | 0 | symptom assessment tool (China) (4 points) |
2010 | Shi | 32 | 38 | (ST 25, ST 36, BL 20, BL 21, BL 23, BL 25 and ST 37) |
38.51±14.65 | 38.68±15.72 | IBS-D | ROME III | 26 | 20 | 4 weeks | 6 | Mixed | China | 0 | 0 | Overall IBS symptom VAS score (10 points) |
2017 | Lembo (1) | 426 | 427 | Eluxadoline 100mg /placebo BID | 44.4±13.9 | 45.8±14.1 | IBS-D | ROME III | 107 | 73 | 12 weeks | 7 | Mixed | United States | 500/859 | 450/808 | IBS-D global symptom score, Bristol Stool Form Scale |
2017 | Lembo (2) | 383 | 382 | Eluxadoline 100mg /placebo BID | 45.7±13.3 | 47.1±13.8 | IBS-D | ROME III | 113 | 62 | 12 weeks | 7 | Mixed | United States | NS | NS | IBS-D global symptom score, Bristol Stool Form Scale |
2013 | DOVE | 163 | 159 | Eluxadoline 100mg /placebo BID | 43.6±10.9 | 44.6±12.5 | IBS-D | ROME III | 46 | 22 | 12 weeks | 7 | Mixed | United States | 73/165 | 78/159 | IBS Global Symptom score, IBS-SSS |
2015 | Liang Zheng | 218 | 209 | Pinaverium 50mg tid/placebo | 36.9±11.8 | 36.6±12.6 | IBS | ROME III | 131 | 71 | 4 weeks | 7 | Mixed | China | 40/218 | 32/209 | Bowel Symptom Scale (10 points), Bristol stool form scale |
1977 | Levy | 30 | 30 | Pinaverium 50mg tid/placebo | NS | NS | IBS | Clinical | 24 | 17 | 2 weeks | 3 | Mixed | French | NS | NS | NS |
1981 | Delmont | 25 | 25 | Pinaverium 50mg tid/placebo | NS | NS | IBS | Clinical | 19 | 17 | 4 weeks | 4 | Mixed | French | NS | NS | NS |
2005 | Lin Chang | 131 | 128 | Alosetron 1mg/Vitamin C BID | 44±12 | 43±12 | IBS-D | ROME I | 69 | 51 | 12 weeks | 7 | Mixed | United States | 86/130 | 65/128 | Average abdominal pain and stool consistency score (5 points) |
2004 | William D.Chey | 279 | 290 | Alosetron 1mg/Vitamin C BID | 46.2±13.5 | 46.9±12.9 | IBS-D | Clinical | 144 | 119 | 48 weeks | 7 | Women | United States | 297/348 | 261/362 | Average abdominal pain and stool consistency score (5 points) |
2004 | Lembo (1) | 147 | 135 | Alosetron 2mg/Vitamin C BID | 48.9±15.5 | 49.4±13.8 | IBS-D | Rome II | 100 | 62 | 12 weeks | 6 | Female | United States | 145/246 | 127/246 | IBS-D global symptom score, Average stool consistency scores (5 points) |
2004 | Lembo (2) | 457 | 219 | Alosetron 2mg/Vitamin C BID | 48.8±14.0 | 48.6±13.6 | IBS-D | Rome II | 320 | 99 | 12 weeks | 6 | Female | United States | NS | NS | IBS-D global symptom score, Average stool consistency scores (5 points) |
2007 | Krause | 177 | 176 | Alosetron 1mg/Vitamin C BID | 43 | 43 | IBS-D | ROME II | 76 | 54 | 12 weeks | 7 | Women | United States | 102/176 | 94/176 | IBS-D global symptom score, Average stool consistency scores (5 points) |
2011 | Lee KJ | 175 | 168 | Ramosetron 5ug Qd/Placebo | 43.4±12.1 | 45±13.1 | IBS-D | Rome III | 65 | 64 | 4 weeks | 3 | Male | Korea | 69/147 | 77/149 | IBS symptoms (5 points), Bristol Stool Form Scale |
2008 | Matsueda (1) | 297 | 104 | Ramosetron 1ug Qd, 5ug Qd, 10ug Qd/Placebo | 40.3±11.8 | 38.4±9.56 | IBS-D | Rome II | 110 | 28 | 12 weeks | 6 | Mixed | Japan | 177/309 | 61/108 | IBS symptoms (5 points), Bristol Stool Form Scale |
2015 | Fukudo S AB | 307 | 102 | Ramosetron 1.25ug Qd, 2.5ug Qd, 5ug Qd/placebo | 40.9±10.6 | 40.2±10.1 | IBS-D | Rome III | 121 | 29 | 12 weeks | 3 | Female | Japan | NS | NS | IBS symptoms (5 points), Bristol Stool Form Scale |
2016 | Fukudo S | 292 | 284 | Ramosetron 2.5ug Qd/Placebo | 41.4±11.8 | 41.5±12.0 | IBS-D | Rome III | 148 | 91 | 12 weeks | 7 | Female | Japan | 154/292 | 118/284 | IBS symptoms (5 points), Bristol Stool Form Scale |
2008 | Matsueda (2) | 263 | 265 | Ramosetron 5ug Qd/Placebo | 40.7±11.21 | 41.8±11.70 | IBS-D | Rome II | 124 | 72 | 12 weeks | 5 | Mixed | Japan | 163/270 | 141/269 | IBS symptoms (5 points), Bristol Stool Form Scale |
2014 | Fukudo S | 147 | 149 | Ramosetron 5ug Qd/Placebo | 40.9±10.6 | 40.2±10.1 | IBS-D | Rome III | 58 | 26 | 12 weeks | 7 | Male | Japan | 13/175 | 6/168 | IBS symptoms (5 points), Bristol Stool Form Scale |
2008 | Lembo | 191 | 197 | Rifaximin 550mg bid/Placebo | NS | NS | IBS-D | Rome II | 100 | 87 | 2 weeks | 4 | Mixed | Multicenter | NS | NS | IBS-D global symptom score, IBS-associated bloating |
2011 | Primentel (1) | 309 | 314 | Rifaximin 550mg tid/Placebo | 46.2±15.0 | 45.5±14.6 | Non-C | Rome II | 126 | 98 | 2 weeks | 7 | Mixed | Multicenter | 264/624 | 296/634 | IBS symptoms (5 points), Bristol Stool Form Scale |
2011 | Primentel (2) | 315 | 320 | Rifaximin 550mg tid/Placebo | 45.9±13.9 | 46.3±14.6 | Non-C | Rome II | 128 | 103 | 2 weeks | 7 | Mixed | Multicenter | NS | NS | IBS symptoms (5 points), Bristol Stool Form Scale |
2017 | Lowe | 43 | 36 | Acupuncture/sham | 42±15 | 43±15 | IBS | Rome I | 23 | 15 | 4 weeks | 7 | Mixed | Canada | 0 | 0 | IBS Symptoms (5 points), SF-36, IBS-36 |
2000 | Catherine Lowe | 28 | 22 | Acupuncture/sham | NS | NS | IBS | NS | 16 | 10 | 4 weeks | 3 | Mixed | Canada | NS | NS | NS |
2009 | Anthony J | 78 | 77 | Acupuncture/sham | 37.5±14.6 | 38.9±14.1 | IBS | Rome II | 32 | 24 | 3 weeks | 6 | Female | United States | 0 | 0 | IBS-Symptom severity scale, IBS-AR, QOL |
2005 | Forbes | 27 | 32 | Acupuncture/sham | 43 | 44.4 | IBS | Rome+Manning | 13 | 10 | 12 weeks | 7 | Mixed | UK | 0 | 0 | global symptom score, Bristol stool scale |
Identification process for eligible trials.
In total, 9712 patients diagnosed with IBS-D/IBS were enrolled in the assessed studies, mean age was between 38 and 46 years, the diagnosis criteria included clinical criteria, ROME I-III, and the treatment duration was from 2 weeks to 48 weeks, mainly between 4 and 12 weeks. The following seven therapeutic methods were included: A: acupuncture; B: eluxadoline; C: pinaverium bromide; D: alosetron; E: ramosetron; F: rifaximin; and G: sham acupuncture; H: placebo (vitamin C, etc.). Documents included 10 articles from China, 9 articles from the United States, 2 papers from France, 2 papers from Canada, 1 paper from United Kingdom, 5 articles from Japan, 1 article from Korea, and 3 articles from multicentre locations. Using the Jadad scale assessment, the overall Jadad score for study quality ranged from 1 to 7, and the median Jadad score was 4 (see Table
Compared with placebo, acupuncture significantly improved the symptoms of diarrhoea-predominant irritable bowel syndrome (OR: 7.7, 95% CI: 3.8-16.0, I2 = 0%) (Figure
The Forest plot of IBS-D treatment of acupuncture compared with other drugs. A: acupuncture; B: eluxadoline; C: pinaverium bromide; D: alosetron E: ramosetron; F: rifaximin; G: sham acupuncture; H: placebo.
The cumulative probability ranking of the results for diarrhoea-predominant irritable bowel syndrome patients is as follows: acupuncture, sham acupuncture, pinaverium, alosetron = eluxadoline, ramosetron, and rifaximin. The probability distribution rankings of eluxadoline were equal, so we chose the probability of the closest top rank as its ranking result. The efficacy of acupuncture was much higher than that of other drugs (P = 0.977), while sham acupuncture had a higher drug efficacy (P = 0.90) than pinaverium bromide (P=0.69), alosetron (P = 0.35), eluxadoline (P= 0.30), ramosetron (P=0.31), and rifaximin (P=0.81) (Figure
The cumulative probability rankings of treatment effect of acupuncture on IBS-D.
Methods/Rankings | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 |
---|---|---|---|---|---|---|---|---|
Acupuncture | 9.774500e-01 | 0.0221666667 | 0.0003166667 | 6.666667e-05 | 0.000000000 | 0.0000000000 | 0.000000000 | 0.000000e+00 |
Eluxadoline | 7.000000e-04 | 0.0183166667 | 0.1130666667 | 3.042167e-01 | 0.260266667 | 0.2448666667 | 0.058383333 | 1.833333e-04 |
Pinaverium Bromide | 0.000000e+00 | 0.0371500000 | 0.6937166667 | 1.282833e-01 | 0.070783333 | 0.0527000000 | 0.016733333 | 6.333333e-04 |
Alosetron | 4.333333e-04 | 0.0141500000 | 0.1021333333 | 3.490500e-01 | 0.304650000 | 0.1983666667 | 0.031200000 | 1.666667e-05 |
Ramosetron | 1.333333e-04 | 0.0056333333 | 0.0452000000 | 1.890667e-01 | 0.312650000 | 0.3764000000 | 0.070916667 | 0.000000e+00 |
Rifaximin | 1.666667e-05 | 0.0005333333 | 0.0039000000 | 1.546667e-02 | 0.042683333 | 0.1198666667 | 0.806500000 | 1.103333e-02 |
Sham Acupuncture | 2.126667e-02 | 0.9020500000 | 0.0416666667 | 1.385000e-02 | 0.008966667 | 0.0076833333 | 0.003983333 | 5.333333e-04 |
Placebo | 0.000000e+00 | 0.0000000000 | 0.0000000000 | 0.000000e+00 | 0.000000000 | 0.0001166667 | 0.012283333 | 9.876000e-01 |
The cumulative probability ranking plot of treatment effect of acupuncture and other drugs on IBS-D.
There were 22 studies that reported side effect data (Table
The cumulative probability rankings of side effect of drugs on IBS-D.
Drugs/Possibility | 1 | 2 | 3 | 4 | 5 | 6 |
---|---|---|---|---|---|---|
Eluxadoline | 0.0089250 | 0.0265875 | 0.082650 | 0.1953500 | 0.2967500 | 0.3897375 |
Pinaverium | 0.1554125 | 0.1458000 | 0.194750 | 0.1840000 | 0.1073250 | 0.2127125 |
Alosetron | 0.3657750 | 0.4054750 | 0.171675 | 0.0457000 | 0.0092500 | 0.0021250 |
Ramosetron | 0.0335625 | 0.1281875 | 0.381825 | 0.3390875 | 0.0883750 | 0.0289625 |
Rifaximin | 0.4363125 | 0.2926750 | 0.149800 | 0.0673875 | 0.0296375 | 0.0241875 |
Placebo | 0.0000125 | 0.0012750 | 0.019300 | 0.1684750 | 0.4686625 | 0.3422750 |
The cumulative probability ranking plot of side effect of drugs on IBS-D.
We compared all of the included studies and drew network diagrams, with the studies incorporated into quality-based displays on a network map (Figure
The network plot of all treatment methods: yellow means the low-quality studies, green means the high-quality studies.
In view of the different acupuncture points selected for each study, we selected the most commonly used acupoints, including ST-25, ST-37, ST-36, SP-6, GV-20, and EX-HN3; the use of these 6 acupoints was 4 times more common than other acupoints (Table
Most commonly used acupoints in our included articles.
Acupoint Number | Frequency | Positions |
---|---|---|
ST-25 | 10 | Abdomen |
ST-37 | 9 | Leg |
ST-36 | 8 | Leg |
SP-6 | 5 | Leg |
GV20 | 5 | Head |
EX-HN3 | 4 | Forehead |
By performing 20,000 convergence iterations, we obtained a Brooks-Gelman-Rubin diagnostic plot, and the track density map was acceptable; based on the node-splitting model, we found all studies in the region beneath the 4th line. We also obtained a cumulative contribution map from the STATA software (Figures
Brooks-Gelman-Rubin diagnostic plot of included studies.
Density plot of included studies.
Node-splitting plot of included studies.
The cumulative contribution plot of IBS-D treatment of acupuncture compared with other drugs. ACU: acupuncture; ELU: eluxadoline; PIN: pinaverium bromide; ALO: alosetron; RAM: ramosetron; RIF: rifaximin; SHAM: sham acupuncture; PLA: placebo.
Using heterogeneity analysis, we found that alosetron and ramosetron had significant heterogeneity; based on the sensitivity analysis, we corrected the OR for alosetron (OR: 1.29, 95% CI: 1.17-1.42) and the OR for placebo and ramosetron (OR: 1.33, 95% CI: 1.22-1.39), and no large directional change occurred even after corrections (Figure
The heterogeneity analysis of included studies. A: acupuncture; B: eluxadoline; C: pinaverium bromide; D: alosetron; E: ramosetron; F: rifaximin; G: sham acupuncture; H: placebo.
The funnel plot shows that all included studies were compared on a pairwise basis, and all the studies were found to be essentially symmetrical, indicating a small publication bias (Figure
The funnel plot of all included studies. ACU: acupuncture; ELU: eluxadoline; PIN: pinaverium bromide; ALO: alosetron; RAM: ramosetron; RIF: rifaximin; SHAM: sham acupuncture; PLA: placebo.
Through NMA, this article found that the effect of acupuncture treatment on diarrhoea-predominant irritable bowel syndrome was better than that of the assessed drugs, with close to no side effects. Previous studies have shown that the effects of acupuncture treatment on diarrhoea-predominant irritable bowel syndrome are still not yet clear, but there are several relevant studies to prove its possible role in treatment. Several studies have confirmed the co-occurrence of IBS and the excessive release of proinflammatory cytokines and insufficiencies in anti-inflammatory cytokine secretion [
However, previous meta-analyses showed no significant benefit of acupuncture compared with sham acupuncture groups in the treatment of IBS. Only a few studies from China have demonstrated the superiority of acupuncture relative to drugs [
However, in the past, most studies conducted a direct comparison between acupuncture and pinaverium bromide. There is no direct comparison between acupuncture and other drugs such as ramosetron, alosetron, rifaximin, and eluxadoline. In the future, direct comparisons can be used to compare differences in efficacy. At the same time, this article found that the evaluation scale used in acupuncture-related research is different from other drugs (only 4 points), which will lead to a bias in the evaluation to a certain extent. In the meantime, the quantity of previous acupuncture research is relatively low, so the conclusions remain to be confirmed; these findings can be verified by increasing the sample size and using multicentre double-blind randomized controlled studies.
This study also shows sham acupuncture for the treatment of IBS-D was more effective than other drugs. Previously, there was a lack of direct comparison between sham acupuncture and oral placebo drugs, our study provides an indirect result between sham acupuncture and oral drug placebo, and there exists some curative effect for IBS-D. Actually, sham acupuncture uses the blunt needle as control, which is the same as the mechanisms of acupressure, a previous comment showed sham acupuncture may be not a good control for experiment group [
This study shows that pinaverium bromide for the treatment of IBS-D was more effective and had fewer side effects than other drugs. Previously, there was a lack of NMA comparing pinaverium bromide and other drugs. A meta-analysis of antispasmodics showed that the pinaverium bromide-induced overall improvement in symptoms of irritable bowel syndrome was 1.55 (CI 95%: 1.33-1.83) and that improvement in abdominal pain was 1.52 (CI%: 1.28-1.80) [
This study shows that alosetron has better efficacy than ramosetron, but with many side effects. Previous studies have shown the occurrence of side effects from alosetron in the treatment of IBS-D (RR = 1.16, 95% CI: 1.08, 1.25) [
The most frequently used acupuncture points for IBS-D were ST-25, ST-37, ST-36, SP-6, GV-20, and EX-HN3. Studies have reported that the electrical stimulation of rat hind limbs at ST-36 bits can significantly improve colonic hypersensitivity [
This study has several advantages and disadvantages. Limitations include the poor quality of some of our studies, the relatively small number of people included, and the fact that some of the studied populations were regional. At the same time, some studies lacked safety records and some results lacked age records, which could have an impact on the results. Meanwhile, the outcome evaluation index used in this study was an overall symptom improvement scale. The drugs used in this study were single drugs. The lack of a combination effect between drugs will have a certain difference from clinical applications.
In summary, this study found that acupuncture may be a good treatment for IBS-D with few side effects, but more research is needed in the future to prove this. Sham acupuncture may be not a good control because of its curative effect for IBS-D. Pinaverium bromide is also a treatment option, as it showed a curative effect with fewer side effects.
The authors declare that they have no conflicts of interest.
Lingping Zhu was responsible for the design of this study. Lingping Zhu and Shasha Ye performed the search of related articles. Yunhui Ma and Zhiqun Shu performed the extraction of data. Lingping Zhu and Yunhui Ma performed the statistical analysis. Lingping Zhu performed the manuscript editing.
Supplementary 1. Detailed searching strategy.