Is Electroacupuncture an Effective and Safe Treatment for Poststroke Depression? An Updated Systematic Review and Meta-Analysis

Objective To observe and compare the efficacy and safety of electroacupuncture and antidepressants in the treatment of poststroke depression (PSD) using a meta-analysis method. Methods The VIP, CNKI, Wanfang, CMB, Embase, PubMed, and Cochrane databases were searched. All randomized controlled trials (RCT) on electroacupuncture treatment of PSD were searched and further screened. Meta-analysis was performed on electroacupuncture and western medicine for PSD to explore the difference in efficacy between electroacupuncture and western medicine for PSD. Results Nineteen RCTs were included in the meta-analysis. Compared with the Western medicine group, the meta-analysis showed no significant changes in Hamilton Depression Scale (HAMD) scores between the electroacupuncture group and the antidepressant group (P > 0.05). The number of adverse events in the electroacupuncture group was less than that in the antidepressant group. Conclusion Compared with antidepressants, electroacupuncture is not less effective in improving depression symptoms in PSD patients with greater safety.


Introduction
Poststroke depression (PSD) is the most common complication of poststroke affective disorder. It has been a major health issue due to its detrimental effects on cognitive function, social activity, and stroke rehabilitation [1]. PSD is the focus of stroke treatment and prevention in China. As one of the traditional therapies in China, electroacupuncture (EA) has been demonstrated to be effective in the treatment of PSD in a couple of clinical studies [2][3][4]. The early intervention with EA was shown to be beneficial for subsyndromal depression, with significantly improved symptoms and quality of life after 6 weeks of treatment [5]. With the development of evidence-based medicine, more and more randomized controlled trials (RCTs) of electroacupuncture therapy come out gradually. However, its imperfect method-ology results in low study quality. Currently, there have been many RCTs on electroacupuncture treatment of PSD that demonstrated that acupuncture treatment of PSD has definite efficacy and fewer side effects. Therefore, this study will figure out the differences in efficacy and safety between electroacupuncture and antidepressant treatment for PSD based on meta-analysis to determine the advantages of electroacupuncture compared with antidepressant treatment.  (1) Define disease: stroke, stroke, depression (post stroke, post cerebral hemorrhage, post cerebral ischemia, depressive disorder, depressive)

Methods
(2) Definition of intervention measures: Electroacupuncture, Electroacupuncture, Electrical acupuncture (3) Definition of study type: controlled clinical trial, randomized controlled trial, randomized trial (4) "OR" is used between two search terms with the same or similar definitions. "AND" is used between multiple search terms with different definitions 2.2. Inclusion Criteria. The following are the inclusion criteria: (1) RCT literatures are in English and Chinese on electroacupuncture treatment of PSD (2) The subjects were patients with PSD who met the diagnostic criteria of stroke and depression  Two researchers independently evaluated and rated the quality of the include literatures. A score of 1 to 3 indicated low quality, while a score of 4 to 7 indicated high quality.

2
BioMed Research International 2.6.1. Heterogeneity Assessment. Heterogeneity in the included studies was assessed by I 2 values. If I 2 < 50%, there was no obvious heterogeneity. If I 2 ≥ 50%, the heterogeneity was higher.

Subgroup and Sensitivity
Analysis. If the heterogeneity was high (I 2 ≥ 50%), studies that might lead to heterogeneity were excluded by sensitivity analysis. If the heterogeneity remained high, factors that may lead to heterogeneity were determined; then subgroup analysis was conducted according to these factors.

Assessment of Publication
Bias. If more than 10 studies were included, funnel plots were used to assess publication bias.
2.9. Statistical Analysis. Meta-analysis was performed using Review Manager V.5.3 software. Continuous variables were presented as the mean difference (MD), whereas dichotomous variables were presented as relative risk ratios (RRs) with 95% confidence intervals (CIs). If there was no significant heterogeneity (I 2 < 50%), a fixed effects model was used. The random effects model was used in case of significant heterogeneity (I 2 ≥ 50%). P < 0:05 indicates that the difference is statistically significant.
3.3. Quality Assessment. All 19 studies were claimed to be randomized, but 5 studies [2,9,12,15,19] did not state the method of random sequence generation. Two studies [7,11] reported the procedure for allocation concealment. Nine studies [1, 3, 5, 7, 10, 11, 13, 16, 17] used the random number      [6,8,14] were randomly grouped according to the order of admission or treatment. Two studies [4,18] used computer software to randomly generate random numbers and groups for grouping. None of the studies blinded the control group patients, and only 8 studies [5-7, 9, 13-15, 19] mentioned blinding of outcome assessment. In two studies [7,11], there are three dropouts. In one study [13], seven patients dropped out. The loss rate of all the studies was less than 5%, which did not affect the statistics of the results, so the above thirteen patients were excluded. The Jadad scores of each study are shown in Table 1, and the assessment of bias risk of each study is shown in Figure 2.
(2) Sensitivity Analysis. HAMD scores at week 4 before and after treatment were recorded in thirteen studies, and the meta-analysis was highly heterogeneous (I 2 = 77%), so sensitivity analysis was conducted. According to the principle proposed by Patsopoulos et al. [20], it was found that two of the studies [8,18] had the greatest influence on heterogeneity which was significantly reduced after the deletion of these two studies. However, compared with the antidepressant group, the HAMD score in the electroacupuncture group did not change significantly (I 2 = 9%, P = 0:36; SMD -0.09, 95% CI: -0.24, 0.06, Figure 4).

Publication Bias Analysis.
The funnel plot of the HAMD score at different endpoints ( Figure 6) showed no asymmetry, indicating that the 19 included studies had no evidence of significant publication bias. A funnel plot of the incidence of AEs was not generated due to the fact that the number of included studies is less than 10.

Discussion
As an important indicator to evaluate the status of depression, accurate measurement of HAMD score is of great significance to the diagnosis and treatment of depression. Currently, the effectiveness of HAMD score has been taken as an outcome indicator by a large number of studies. All studies included in this meta-analysis used changes in HAMD score as the primary outcome indicator of results. All included studies used the HAMD scale to assess depressive symptoms. We evaluated the improvement of depressive symptoms after 4, 6, and 8 weeks of treatment. Among them, HAMD scores were evaluated at week 6 and week 8 after treatment. Meta-analysis showed that there were no significant changes of the HAMD score in the EA group compared to the antidepressant group. Thirteen studies recorded HAMD scores at week 4 before and after treatment. Metaanalysis showed a significant reduction in HAMD scores in the EA group compared to the antidepressant group, but the results showed increased heterogeneity. After using sensitivity analysis to remove the two studies with the greatest impact, it was found that the HAMD score was still not sig-nificantly changed in the EA group compared with the antidepressant group. Adverse events were reported in nine of the 19 studies included. Meta-analysis showed fewer adverse events in the EA group than in the antidepressant group, and there was no significant heterogeneity. Among them, one study [12] adopted the TESS score, revealing that the score of the EA group changed significantly compared to that of the antidepressant group. In summary, EA are effective in improving depressive symptoms. Compared with antidepressant medicine, EA has the advantages of fewer side events and better curative effects. In terms of safety, EA are superior to antidepressants. Jadad ratings are displayed. Five studies [4, 5, 7, 10, 11] scored 3. Seven studies [1,3,13,14,[16][17][18] scored 2. However, the rest of the studies [2,6,8,9,12,15,19] scored only 1. The quality evaluation of the included literature identified the following problems: (1) the random method was not rigorous enough or even missing: among all the included studies, nine studies used the randomized numerical tables. Five studies did not mention random sequence generation methods. The randomization principle in three studies was the order of visits. In two studies, computer software was