An Overview of Systematic Reviews and Meta-Analyses on Acupuncture for Post-Acute Stroke Dysphagia

Background: Many randomized controlled trials (RCTs) and systematic reviews (SRs) on acupuncture treatment for post-acute stroke dysphagia have been published. Conflicting results from different SRs necessitated an overview to summarize and assess the quality of this evidence to determine whether acupuncture is effective for this condition. The aim was to evaluate methodological quality and summarizing the evidence for important outcomes. Methods: Seven databases were searched for SRs and/or meta-analysis of RCTs and quasi-RCTs on acupuncture for post-acute stroke dysphagia. Two authors independently identified SRs and meta-analyses, collected data to assess the quality of included SRs and meta analyses according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and the revised Assessment of Multiple Systematic Reviews (AMSTAR 2). Results: Searches yielded 382 SRs, 31 were included. The quality of 22 SRs was critically low, five SRs were low, and four Cochrane SRs were moderate when evaluated by AMSTAR2. A total of 17 SRs reported 85.2–96.3% of PRISMA items. Five SRs included explanatory RCTs, 16 SRs included pragmatic RCTs, and 10 SRs included both. Conclusion: Currently, evidence on the effectiveness of acupuncture on post-acute stroke dysphagia is of a low quality. The type of study appeared to have no direct influence on the result, but the primary outcome measures showed a relationship with the quality of SRs. High quality trials with large sample sizes should be the focus of future research.


Introduction
Stroke is considered to be one of the leading causes of adult mortality and disability [1], and the lifetime risk of stroke occurs in approximately 25% of adults over 25 years old [2]. Dysphagia is a common complication that occurs in 37% to 78% of stroke survivors who experience various problems such as eating slowly, having difficulty swallowing when drinking water, and these issues are often accompanied with a speech disorder [3]. Additionally, pneumonia, chest infection or even death may occur as a result of dysphagia [3,4]. Due to the lack of direct treatment available for post-acute stroke dysphagia (surgery or medicine), early screening for these patients is recommended to prevent the

Selection of the Systematic Reviews and Meta Analyses
The search identified 382 articles, including 12 from PubMed, 60 from VIP, 31 from EMBASE, 15 from the Cochrane Library, 31 from CNKI, 111 from Sino-Med and 122 from the Wanfang Database. After 153 duplicate records were removed, the titles and abstracts of the remaining 229 records were reviewed. Of these, 189 records were removed after screening titles and abstracts as they were irrelevant, seven SRs were not relevant to post-acute stroke dysphagia, 15 SRs were not relevant to acupuncture and a further eight SRs were duplications. A total of 40 potentially relevant articles were downloaded for full-text screening. Subsequently nine studies were excluded, and the reasons for exclusion were as follows: one SR included observational studies, two articles were conference abstracts, one SR we were unable to get the full text, four studies were not SRs, and one article was a duplicate. Finally, 31 studies were included in our study [8][9][10]. The study flow chart is shown in Figure 1.
A total of 10 SRs [10,23,[30][31][32][33][35][36][37]40] mentioned adverse events (fewer than 3% occurred in each SR), while only five SRs [10,31,36,37,40] reported adverse events that were associated with acupuncture, and four SRs [23,30,32,35] reported pain, ecchymosis and hematoma during the process or at the site of needling, while the other SR [33] reported the pain occurred after electro acupuncture. These data may suggest that acupuncture rarely caused serious side effects, and that acupuncture could be considered as safe for the treatment of post-acute stroke dysphagia. All included SRs concluded that there was very low to low quality evidence on the effectiveness of acupuncture treatment on post-acute stroke dysphagia, there is still a need for high quality trials with large sample sizes. Table 1 lists the characteristics of these included SRs. The therapeutic effect of acupoint stimulating therapy on post stroke dysphagia is better than that the control group, but more well designed randomized, are needed to support this conclusion. Acupuncture combined with rehabilitation training is effective for post stroke dysphagia and the combined effect is better than rehabilitation training alone. However, due to the small size and low quality of included RCTs, well designed RCTs with large-scale and high-quality are still required. Acupuncture combined with swallowing training has obvious effect for post stroke dysphagia. The swallowing function of patients improved more obviously than that of the control group at the same time. 16  A reliable conclusion cannot be drawn from the present data because of the low methodological quality, especially because of the lack of data on long-term outcomes. A tendency that acupuncture can improve dysphagia after stroke in short-term with no adverse effect id demonstrated. Therefore, it is necessary to conduct more multi-central RCTs with high quality in future. 18  Conventional treatment plus acupuncture was more effective for some outcomes than conventional treatment alone, but RCTs with higher quality in the future may produce new evidence. Acupuncture combined with drugs is better than simple drugs assessed by VFSS scores. Acupuncture combined with drugs and rehabilitation training is better than drugs combined with rehabilitation assessed by SSA and VFSS scores. Acupuncture has a positive effect on improving the WST, SSA and VFSS score. However, it has not been proven that acupuncture combined with drugs and rehabilitation training can reduce the incidence of aspiration pneumonia. Acupuncture combined with rehabilitation is better than rehabilitation training alone but acupuncture and rehabilitation training have the same effect on the treatment of patients with post stroke dysphagia. However, the long-term effect of acupuncture on post stroke dysphagia is better than rehabilitation training. 23   Definitive conclusions on acupuncture with conventional rehabilitation therapy for post stroke dysphagia cannot be made due to the low-quality evidence, but this combination approach appears to be promising. We recommend that acupuncture may still be used as combination use by qualified practitioners as it is relatively safe without much negative effect 27  Death or dependency at the end of follow-up GRADE From the available evidence, acupuncture may have beneficial effects on improving dependency, global neurological deficiency, and some specific neurological impairments for people with stroke in the convalescent stage, with no obvious serious adverse events. However, most included trials were of inadequate quality and size.

Acupuncture Versus Rehabilitation Training/Western Medicine/Routine Therapy
Three SRs [20,25,40] compared acupuncture alone with rehabilitation training/conventional therapy, one SR [20] showed that ER and overall recovery rate in acupuncture group were higher than rehabilitation training or routine therapy (relative risk (RR) = 1. 38 [41] just included one trial of 66 participants and demonstrated that the acupuncture group showed no statistically significant differences when compared with baseline group. The relative risk of recovery was 1.61 with a 95% CI of 0.73 to 3.58. One SR [19] compared different stage of stroke, for stroke in the convalescent phase, the ER for acupuncture plus baseline treatment group was higher than the baseline control group [RR = 1.45; 95% CI [1.16, 1.80]; p = 0.001; eight studies; 766 participants), but there was no difference at the acute stage.

Discussion
Acupuncture is used as routine clinical therapy in China for post-acute stroke dysphagia. There is some evidence that it can improve the cerebral blood flow and serum levels of brain-derived neurotrophic factor (BDNF) and nerve growth factor (NGF) in dysphagia patients [42], but still there is a lack of widely agreed evidence of a biologically-plausible basis for its effect. This overview summarizes the current evidence on the effectiveness of acupuncture for post-acute stroke dysphagia. Of the 31SRs and meta-analyses identified, most were of critically low quality assessed by AMSTAR2, and almost half of them reported 85.2-96.3% items on PRISMA. Due to the very low to low quality of evidence, and insufficient reporting data provided by these SRs and meta-analyses, there is still no definitive conclusion on the effectiveness of acupuncture for post-acute stroke dysphagia.
Most SRs and meta-analyses included in this overview ignored the need to register the protocol, and no SRs and meta-analyses provided an explanation for including only RCTs, future studies should pay attention to these two items in AMSTAR 2. The included 31 SRs and meta analyses in this overview were published between 2006 and 2019, nine SRs [9,17,20,21,26,30,38,39,41] were published before 2012 and seven SRs [10,[22][23][24]34,36,40] were published between 2018 and 2019. However PRISMA was published in 2009 [13] and the AMSTAR 2 was updated in 2017 [12]. Many journals require authors to self-evaluate according to the PRISMA statement when submitting a systematic review and meta-analysis. This may be one of the reasons that most of the SRs were critically low as assessed by AMSTAR 2. In addition, many SRs included trials that were conducted prior to the development and use of these two quality assessment tools. As 15 SRs [8,[14][15][16]18,19,25,[27][28][29][31][32][33]35,37] were published between 2013 and 2017, the increasing number of SRs focusing on acupuncture for post-acute stroke dysphagia not only indicates the interest and concern regarding effectiveness of acupuncture in this area, but it also means that SRs are widely used to assess the therapeutic effect based on the original studies. However, the quality of current SRs are low. As we know, AMSTAR is a quality assessment tool used just for SRs of RCTs, while AMSTAR 2 is an update of AMSTAR, which can be used to appraise SRs of intervention trials including both RCTs and NRCTs [12]. We can see from the 16 items of AMSTAR 2 and 27 items of PRISMA, that some items are mutually complementary. PRISMA emphasizes the structure of the SR, while AMSTAR 2 is concerned more with the details of methodology used for included original studies in the SR, especially the risk of bias (including additional bias). Item 27 in the PRISMA checklist stresses the importance of funding for the SR and availability of other support (e.g., supply of data), while AMSTAR 2 emphasizes the funding source of original studies, both of quality measures take potential conflict of interest into account, but only AMSTAR 2 specifically lists the item of conflict of interest. Therefore, the combined use of the two tools can provide an overall assessment of the quality of SRs that focus on healthcare interventions. Future SRs should conduct and report the SRs according to these two tools.
The overall quality of the four Cochrane SRs [8][9][10]41] were assessed as moderate by AMSTAR 2, and these SRs reported more than 90% items of PRISMA. But the results of these four SRs were still inconsistent. One of the Cochrane SRs [41] only included one trial of 66 participants, and compared a routine treatment combined with acupuncture with routine treatment alone, but the statistical significance regarding the primary outcome of feeding was not reported, the relative risk (RR) of recovery was 1.61 with a 95% confidence interval (CI) [0.73, 3.58]. One of the SRs [8] which included three RCTs showed that acupuncture was superior to no acupuncture in the terms of improving swallowing function as measured by the drinking test (mean difference (MD) = −1.11, 95% CI [−2.08, −0.14]; participants = 200; studies = 2; I 2 = 96%), and the difference of another included study was also significant (odd ratio (OR) = 95.29, 95% CI [10.93, 830.86]). However, the opposite result was reported in 2012 [9], comparing to sham acupuncture or no acupuncture, there was no difference in swallowing scores between treatment and acupuncture groups. But the heterogeneity was significant (t = 3; n = 256; MD = −0.41; 95% CI [−1.53, 0.72]; I2 = 91%; p < 0.0001), and the updated Cochrane SR [10] supported the result that acupuncture did not improve swallowing ability (SMD −0.55, 95% CI −1.20 to 0.11; 496 participants; six studies; I 2 = 91%; p = 0.10), still with significant heterogeneity, but acupuncture can reduce the number of participants with dysphagia at end of trial (OR = 0.31, 95% CI [0.20, 0.49]; 676 participants; eight studies; I 2 = 0%; p < 0.00001).Therefore, the quality of included RCTs was also another reason that influenced the quality of SRs.
Apart from the methodological problems of the included RCTs, there was diversity in the selection of primary outcomes. Effective rate (ER) and water swallow test were the most frequently used measures to assess the effectiveness of acupuncture on post-acute stroke dysphagia, but the concept of ER was different among these studies, and most studies did not clarify the definition of ER. Just six SRs [22][23][24]28,38,40] reported the definition of ER: effective rate= ('recovery' + 'markedly improved' + 'improved')/total number of patients, and 'recovery' meant totally cure, 'markedly improved' represented nearly complete resolution of dysphagia, while 'improved' represented partial resolution of dysphagia. One SR [40] used the change in water swallow score to assess the resolution of dysphagia, but the authors of this SR did not report how they assessed the resolution of dysphagia. This may have overestimated the efficacy of the intervention to some extent, and future research should provide clear definitions of related outcomes. Besides, just three SRs [8,19,41] mentioned the different stages of stroke, almost each SR included both cerebral hemorrhage and infarction without subgroup analysis and only one SR [40] used CONSORT and STRICTA checklist to evaluate the included RCTs, these studies provide insufficient information on what and how acupuncture was delivered. These are important factors that may influence the efficacy of acupuncture but data was unavailable from the original RCTs.
Due to some of the same primary studies being included in different SRs, and updates of SRs and meta-analyses, data synthesis was not appropriate. Although, the method used for this overview was performed according to the criteria for conducting overviews of SRs and meta analyses given in the Cochrane Handbook of Systematic Reviews of Interventions [43], it still has some limitations, as there are no clear standards for conducting an overview of systematic reviews and meta-analyses. Although a comprehensive literature search was conducted, relevant SRs may have been missed. Stroke was defined as a Mesh word, and acupuncture, dysphagia were used as key words in the title and abstract. This overview may have missed some stroke relevant studies that did not list the acupuncture as treatment or dysphagia as a symptom in the title or abstract.

Conclusions
Currently there is very low to low quality evidence on the effectiveness of acupuncture for post-acute stroke dysphagia. RCTs with high quality and large sample sizes are needed as well as SRs and meta-analyses with high quality. Although the evidence was insufficient to provide definitive conclusions on the effectiveness of acupuncture for post-acute stroke dysphagia, there are preliminary indications that it may improve symptoms associated with dysphagia. Future SRs should consider not only the reporting quality but also the methodological quality when conducting a SR, some details like the clear definition of primary outcomes, the subtype and the different stage of disease should also be considered.
Author Contributions: Thanks to all authors in this study. N.R. conceptualized the study, X.L. and N.R. designed the study and organized the team.