Efficacy and Safety of Chaihu Jia Longgu Muli Decoction in the Treatment of Poststroke Depression: A Systematic Review and Meta-Analysis

Objective Chaihu Jia Longgu Muli decoction (CLMD) is widely used in the treatment of poststroke depression (PSD) in China. Some evidences show that it has advantages, but there lacks reliable evidence. This study aims to systematically evaluate the efficacy and safety of CLMD in the treatment of PSD. Methods All randomized controlled trials (RCTs) of CLMD in the treatment of PSD were searched from the following databases: PubMed, Cochrane Library, Embase, Web of Science, China National Knowledge Infrastructure (CNKI), Wanfang Database, VIP Database, and Chinese Biomedical Literature Service System (CBM), from their inception to May 2021. Two researchers independently screened the literature, extracted the data, and evaluated the risk of bias in the included studies. Meta-analysis was performed using RevMan5.3 software. Results A total of 13 RCTs involving 1665 patients were finally included in this study, among which 5 RCTs were oral CLMD alone versus antidepressants, and 8 RCTs were oral CLMD with antidepressants versus antidepressants. Meta-analysis results showed that oral administration of CLMD could improve Hamilton's Depression Scale (HAMD) and the Modified Edinburgh-Scandinavian Stroke Scale (MESSS) scores, improve the Barthel index, and have a low rate of adverse reactions, but there was no significant difference in the total effective rate (p=0.21 > 0.05) and the National Institute of Health Stroke Scale (NIHSS) score (p=0.47 > 0.05) between the antidepressants group and the oral administration of the CLMD group. Oral CLMD combined with antidepressants could improve the total effective rate, HAMD, and MESSS score, but there was no significant difference in Barthel index (p=0.06 > 0.05) and the adverse reaction rate (p=0.14 > 0.05) between the two groups. Conclusion Current evidence suggests that oral CLMD alone or with antidepressants is more effective and safer in the treatment of PSD than oral antidepressants. Due to the limitation of the quality and quantity of the included studies, more high-quality studies are needed to confirm the above conclusion.


Introduction
Poststroke depression (PSD) refers to a series of psychological and physical syndromes featured with depression, slow response, loss of interest, and other symptoms after stroke [1]. e incidence of PSD ranges from 29% to 31% [2], and it usually occurs within 1 year after stroke [3]. PSD is closely related to the poor prognosis of stroke, which leads to prolonged hospitalization, neurological recovery disorder, more loss of independent living ability, and even increased mortality [1,4,5]. Studies have shown that the mortality of patients with PSD is significantly higher than that of patients with stroke alone, which is 1.28-1.75 times higher, and the severity of depression is highly correlated with the mortality [6]. Antidepressants are the first choice for the treatment of PSD, including the selective serotonin reuptake inhibitor (SSRI), serotonin norepinephrine 2. Methods e protocol and registration information are available at https://www.crd.york.ac.uk/prospero/display_record.php? ID�CRD42021255407 (registration number: CRD4202125 5407). We performed this meta-analysis according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Table S1). e retrieval method adopted the combination of medical subject headings (MeSH) terms and free terms, and the English retrieval words mainly included stroke, cerebrovascular accident, depression, Chaihu Jia Longgu Muli decoction , and Chaihu Jia Longgu Muli granules (Table S2). All literature were reviewed by two investigators (Renhong Wan and Yihua Fan) independently. Any disagreement was resolved by the consultation with a third researcher (Ruiwen Song).

Inclusion and Exclusion Criteria
2.2.1. Inclusion Criteria. Inclusion criteria were as follows: (1) Study type: randomized controlled trials (RCTs) of CLMD in the treatment of PSD (2) Diagnostic criteria: the diagnostic criteria for stroke refer to the Diagnostic Points of Various Cerebrovascular Diseases [16], for depression refer to the ones for PSD in the Classification and Diagnostic Criteria of Mental Disorders in China [17], and for TCM refer to the Diagnostic Efficacy Criteria for Diseases of TCM (3) Interventions: the treatment group was given CLMD or combined with the antidepressant, while the control group received the same antidepressant as the treatment group. (Oral preparations of Chaihu Jia Longgu Muli (CLM) included different forms such as CLMD and CLM granules. Modified CLMD referred to the addition or subtraction of no more than 3 herbs in PSD patients with different symptoms [18]). (4) Outcome indicators: the main outcome indicators are (i) total effective rate: efficacy was assessed by the reduction rate of Hamilton's Depression Scale (HAMD) score. e criteria for the efficacy of depression were recovery, score reduction rate >75%; significant effect, the reduction rate >50%; effectiveness, score reduction rate ≥25%; ineffectiveness, subtraction rate <25%; and total effective rate-� (recovery number + significant effect number + effectiveness number)/total number * 100%; and (ii) HAMD score: secondary outcome indicators: (i) Barthel index for activities of daily living; (ii) National Institute of Health Stroke Scale (NIHSS); (iii) the Modified Edinburgh-Scandinavian Stroke Scale (MESSS); and (iv) adverse reactions rate.

Exclusion
Criteria. Exclusion criteria were as follows: (1) For repetitive studies, only the studies with the highest quality and best data were included (2) Studies with incomplete data or significant errors that cannot be resolved after contact with the author (3) Studies in which random methods or allocation concealment were evaluated as high risk (4) Studies without outcome indicators Two independent  reviewers screened the literature strictly according to the  inclusion and exclusion criteria, extracted the data, and  cross-checked the data after completion. For literature and  data with objections, two reviewers would discuss, and if  there was no agreement, the third reviewer would be invited  to evaluate. e extracted data included (i). basic information of the included studies: title, first author, year of publication, number of cases in each group, and baseline characteristics of patients; (ii) intervention measures and treatment course of treatment group and control group; (iii) outcome indicators; and (iv) each risk bias assessment elements in RCTs.

Risk of Bias Assessment.
e risk of bias in the included studies was independently evaluated by two investigators, and the results were cross-checked. If there was any disagreement, it would be discussed and resolved with the third researcher. Bias risk assessment adopted the RCT bias risk assessment tool recommended in the Cochrane manual 5.1.0 [19].

Statistical Analysis and Data Synthesis for Meta-Analysis.
RevMan5.3 software recommended by the Cochrane Collaboration was used for meta-analysis. For continuous variables, if the measurement tools and units were the same, the weighted mean difference (WMD) was used for analysis. If the measurement tools or units were inconsistent, the standard mean difference (SMD) was used. e dichotomous variables were analyzed by relative risk (RR), and 95% confidence interval (95% CI) was used for each effect size. e heterogeneity among the included results was analyzed by the χ 2 test, and the magnitude of the heterogeneity was determined quantitatively by combining with I 2 . If p > 0.10 and I 2 < 50%, there was no significant heterogeneity between studies, and a fixed-effect model was used for meta-analysis. If p > 0.10 and I 2 ≥ 50%, the heterogeneity between studies was considered significant, and then, subgroup analysis or sensitivity analysis was used to explore the source of heterogeneity. After the exclusion of obvious clinical heterogeneity and methodological heterogeneity, a random-effect model was used for metaanalysis. Sensitivity analysis was used to observe the influence of single study on the combined effect size and to analyze the stability of the meta-analysis results. For main outcome indicators, if the included studies were ≥10, the funnel plot was used to qualitatively detect publication bias. Egger's and Begg's tests were used to quantitatively assess the potential publication bias.

Characteristics of the Studies.
Of 277 related articles obtained by the initial search, 102 were obtained after removing the duplicates. After reading the titles and abstracts, 55 articles were excluded, and 34 were excluded after reviewing the full text in the remaining 47, so 13 studies were eligible for inclusion. e literature screening process is shown in Figure 1. e basic characteristics of the included studies are given in Table 1.
e treatment group was treated with CLMD alone or combined with antidepressants, while the control group was treated with the antidepressants, and the characteristics of the intervention measures are given in Table 2.

Oral CLMD Alone vs. Antidepressant
(1) Total Effective Rate. 4 RCTs reported the total effective rate. e heterogeneity test showed no statistical heterogeneity (p � 0.97; I 2 � 0%). Meta-analysis of the data using a fixed-effect model showed no statistically significant difference between the treatment group and the control group ((RR � 1.05, 95% CI: 0.97, 1.15, p � 0.21 >0.05), Figure 3). Due to the large span of treatment courses in each study, we performed a subgroup analysis based on the course. According to the course of treatment, they were divided into two subgroups: <30 days and ≥60 days. e heterogeneity test of total effective rate showed no statistical heterogeneity in <30 days (p � 0.88; I 2 � 0%) and ≥60 days (p � 0.88; I 2 � 0%) ( Figure 4). Meta-analysis of data using a fixed-effect model indicated that there was no statistical difference in the two groups (<30 days (RR � 1.04, 95% CI: 0.95, 1.14, p � 0.40 >0.05) and ≥60 days (RR � 1.08, 95% CI: 0.92, 1.27, p � 0.35 >0.05)).
(2) HAMD Score. 4 RCTs reported the HAMD score. e heterogeneity test showed no statistical heterogeneity (p � 0.64; I 2 � 0%). Meta-analysis of the data using a fixedeffect model showed that the score of the treatment group was lower than that of the control group, and the difference was statistically significant ((MD � −1.30, 95% CI: −1.99, −0.61, p � 0.002 <0.05), Figure 5). At the same time, we performed subgroup analysis according to the course of treatment.
e heterogeneity test of the HAMD score showed no statistical heterogeneity in <30 days (p � 0.38; I 2 � 0%) and ≥60 days (p � 0.87; I 2 � 0%) ( Figure 6). Meta-Evidence-Based Complementary and Alternative Medicine analysis of data was performed using a fixed-effect model, and the results indicated that the HAMD score of the treatment group was better than that of the control group after treatment in the <30 days subgroup (MD � −1.57, 95% CI: −2.36, −0.78, p < 0.0007), but there was no statistically significant difference between the two groups in the ≥60 days subgroup (MD � −0.66, 95% CI: −2.07, 0.76, p � 0.36 >0.05).
(3) MESSS Score. 1 RCT reported the MESSS score, which could not be used for meta-analysis. Descriptive analysis showed that the treatment group was superior to the control group, and the difference between the two groups was statistically significant ((MD � −5.72, 95% CI: −8.05, −3.39), p < 0.00001).

(5) Barthel Index. 3 RCTs reported Barthel index.
e heterogeneity test suggested significant heterogeneity (p � 0.004; I 2 � 82%), so the elimination method was adopted one by one. When Liu's study [14] was excluded, heterogeneity disappeared (p � 1.00; I 2 � 0%, Figure 7), suggesting that this study was the source of heterogeneity, and a fixed-effect model was adopted after the exclusion of heterogeneity. Results showed that the score of the treatment group was better than that of the control group, and the   Evidence-Based Complementary and Alternative Medicine difference was statistically significant (MD � 9.00, 95% CI: 6.45, 11.55, p < 0.00001).
(6) Adverse Reactions Rate. 2 RCTs reported adverse reactions rate. e heterogeneity test showed no statistical heterogeneity (p � 0.59; I 2 � 0%, Figure 8). Meta-analysis of the data using a fixed-effect model showed that the adverse reactions rate in the treatment group was lower than that in the control group, and the difference was statistically significant (RR � 0.10,95% CI: 0.01, 0.75, p � 0.03 <0.05).
e heterogeneity test showed no statistical heterogeneity (p � 0.84; I 2 � 0%). Meta-analysis of the data using a fixed-effect model showed that the total effective rate of the treatment group was better than that of the control group, and the difference between the two groups was statistically significant ((RR � 1.28, 95% CI: 1.19, 1.39, p < 0.00001), Figure 9). We also conducted subgroup analysis according to the course of treatment, and the

Evidence-Based Complementary and Alternative Medicine
(2) HAMD Score. 7 RCTs reported the HAMD score. e heterogeneity test indicated significant heterogeneity (p < 0.00001; I 2 � 99%). However, the confidence intervals in the forest plot were all on the left side of the invalid line, indicating that the heterogeneity among studies did not affect the results. erefore, a random-effect model was adopted for the combination. e results showed that the HAMD score of the treatment group was better than that of the control group, and the difference was statistically significant ((MD � −5.64, 95% CI: −10.11, −1.16, p � 0.01 <0.05), Figure 11). To explore the source of heterogeneity, we performed subgroup analysis according to the course of treatment. e heterogeneity test result of the subgroup (≥30 and <60 days) was p < 0.00001 and I 2 � 99%, indicating significant heterogeneity. When Lai's study was excluded [24], the heterogeneity disappeared (p � 0.41; I 2 � 0%), suggesting that this study was the source of heterogeneity. A fixed-effect model was used for meta-analysis of the data,    (3) MESSS Score. 1 RCT reported the MESSS score, which could not be used for meta-analysis. Descriptive analysis showed that the treatment group was superior to the control group, and the difference between the two groups was statistically significant (MD � −5.26, 95% CI: −7.55, −2.97, p < 0.00001).
(4) NIHSS Score. 4 RCTs reported the NIHSS score. e heterogeneity test indicated significant heterogeneity (p < 0.00001; I 2 � 99%). A one-by-one elimination method was used to analyze the source of heterogeneity. When Liu and Zhang's study [26] was excluded, the heterogeneity was significantly reduced (p � 0.19; I 2 � 37%, Figure 13), suggesting that the study was the source of heterogeneity, and a fixed-effect model was adopted after the exclusion of heterogeneity. Results showed that the NIHSS score of the treatment group was better than that of the control group, and the difference was statistically significant (MD � −2.93, 95% CI: −3.39, −2.47, p < 0.00001).
(5) Barthel Index. 2 RCTs reported Barthel index, and the heterogeneity test suggested significant heterogeneity (p � 0.03; I 2 � 80%). Due to the small number of the included studies that could not be further analyzed and the study results were all on the side of the invalid line, a random-effect model was adopted. e result showed that there was no statistically significant difference between the two groups ((MD � 8.23, 95% CI: −0.41, −16.87, p � 0.06 >0.05), Figure 14). e heterogeneity test showed no statistical heterogeneity (p � 0.98; I 2 � 0%, Figure 15). Metaanalysis of the data using a fixed-effect model showed that there was no significant difference between the two groups (RR � 0.65, 95% CI: 0.37, 1.16, p � 0.15 >0.05).

Sensitivity
Analysis. Sensitivity analysis of the above indicators was conducted by the one-by-one elimination method, and changes of the effect size and p value were observed after the one-by-one exclusion of the included studies. e results showed that the effect size of outcome indicators did not change significantly, suggesting that the results of the meta-analysis were reliable and stable.

Publication Bias.
Egger's test and Begg's test were used to evaluate whether there was publication bias in the main outcome indicators. For oral CLMD alone vs. antidepressant, no evidence of publication bias was found in the effective rate (Egger's test p � 0.7165 >0.05, Begg's test p � 0.382 >0.05) as well as the HAMD score (Egger's test p � 0.6926 >0.05, Begg's test p � 1.6918 >0.05). As for oral CLMD + antidepressant vs. antidepressant, there was publication bias in the effective rate (Egger's test p � 0.002 <0.05, Begg's test p � 0.0354 <0.05) and the HAMD score (Egger's test p < 0.0001, Begg's test p � 1.9285 >0.05).

Discussion
In this study, a meta-analysis of 13 RCTs of CLMD in the treatment of PSD showed that the following. (1) In terms of total effective rate, we found that CLMD combined with antidepressants was more effective than antidepressants alone, while there was no difference between CLMD and antidepressants alone; (2) HAMD is the most commonly used in the assessment of depressive symptoms, and both CLMD alone and CLMD with antidepressants were better than antidepressants alone in reducing HAMD scores. Depending on the course of treatment, we found different conclusions. When the course of treatment was <30 days, oral CLMD was more effective than antidepressants alone. When the course of treatment was ≥60 days, the efficacy of oral CLMD was comparable to that of antidepressants alone. Whether the treatment course was short or long, the efficacy of CLMD combined with antidepressants was better than that of the antidepressants group, indicating that the treatment of CLMD combined with antidepressants was more conducive to improving the clinical efficacy; (3) MESSS and NIHSS are the international major indicators for the evaluation of neurological function recovery after stroke, which are of great significance for the judgment of    [31]. e higher the score is, the lower the BI value is [32]. Due to the limited number of the included studies, only the NIHSS scores of oral CLMD combined with antidepressants were meta-analyzed, and the results showed that the combined treatment group was superior to the antidepressant group; (4) Barthel index is an indicator to test the independent living ability of patients, which can reflect the degree of nursing need of the patients. Barthel index also can be used to evaluate the functional recovery of PSD patients. As the score of specific items of Barthel index was not reported in the included studies, it was impossible to objectively evaluate the specific impact of CLMD alone or CLMD + antidepressant on the    independent living activities of PSD patients. e results showed that the BI score of oral CLMD alone was better than that of the antidepressant group, indicating that CLMD was positive and effective in improving the independent living ability of PSD patients, but there was no significant difference between oral CLMD combined with antidepressants and the antidepressant group, which may be related to the course of treatment; (5) in terms of adverse reactions, there were a total of 5 RCTs in our study that described adverse reactions, among which two RCTs were about CLMD alone vs. antidepressant, which reported that no adverse reactions occurred in the treatment group, while the adverse reactions in the control group included insomnia, gastrointestinal discomfort, dizziness, and headache.
e remaining three RCTs were related to CLMD + antidepressant vs. antidepressant, among which one RCT reported no adverse reactions in the treatment group and the control group, while two RCTs reported no significant differences in adverse reactions between the treatment group and the control group, including abnormal blood and urine routine, abnormal liver function, insomnia, and digestive tract discomfort. It is seen that CLMD does not increase the risk of adverse reactions, but it does not reduce the side effects of depression; and (6) we conducted sensitivity analysis on all outcome indicators through the one-by-one exclusion method, and the results showed that our meta-analysis was robust.
Stroke is an important social psychological factor leading to depression. Neurological dysfunction and long-term disability caused by stroke lead to the psychological stress response, which brings about psychological imbalance [1]. Depression hinders the recovery of the neurological function after stroke. Antidepressant treatment can not only relieve the symptoms of depression but also promote the physical recovery of stroke patients, which is far more vital than the treatment of depression itself [33].
After thousands of years of exploration, Chinese medicine has advantages in the treatment of mental disorders [34][35][36]. CLMD is one of the most common prescriptions used in the treatment of mental diseases in TCM. Clinical studies have found that CLMD has the effect of psychotropic drugs and have shown a significant antidepressant effect on animal models [37,38]. It can regulate the hypothalamopituitary-adrenal system dysfunction by preventing the dopaminergic and serotonergic transmission in the prefrontal cortex [39] and upregulate the expression of the brain-derived neurotrophic factor (BDNF) to alleviate the depression-like state induced by chronic stress [37]. It also has immediate and long-lasting antidepressant effects by enhancing BDNF expression in the hippocampus [38]. Chaihu (Radix Bupleuri) and Huang Qin (Radix Scutellariae) are the key drugs in many prescriptions for mental disorders. Modern pharmacological studies have found that they can reduce neuroinflammation [40] and neuronal apoptosis [41] and increase the concentration of the nerve growth factor and BDNF [42]. Baicalin in Huang Qin (Radix Scutellariae) can inhibit inflammation [43] and promote nerve regeneration [44,45]. Baicalin also has an antidepressant-like effect [46], which is associated with the increase of BDNF in the hippocampal region [44].
Limitations of this study are as follows: some of the included studies seldom describe the specific operation of the allocation concealment and blind method, and there may be selectivity bias and measurement bias; all the studies are from China, and there may be regional restrictions; due to the particularity of TCM decoction, the composition and dosage of CLMD in the study were different, which may have a certain influence on the results of the study.

Conclusion
Current evidence supports the efficacy of CLMD in PSD patients, which can not only improve depressive symptoms but also promote the recovery of neurological and limbs functions in stroke patients. e efficacy of CLMD alone is no less than that of antidepressants, and there are fewer adverse reactions. In addition, CLMD alone was more effective when treatment was less than 30 days, while oral CLMD combined with antidepressants was more effective than antidepressants alone in both short-and long-term treatment.