A literature search was conducted using the following databases up to 31 January 2023: Embase, Medline, Allied and Complementary Medicine Database, and Cochrane Central Register of Controlled Trials for global databases; Korean Medical Database, KoreaMed, Korean Studies Information Service System, National Digital Science Library, Oriental Medicine Advanced Searching Integrated System, China National Knowledge Infrastructure Database, and Citation Information by Nii for Asian databases.

Symptom-related keywords such as “dyspepsia,” “indigestion,” “pain,” “distress,” and “discomfort” and intervention-related keywords such as “Soyo,” “Xiaoyao,” and “Shoyo” were merged to form a search formula (Table 1). The detailed searching strategies for all databases were presented in the Supplementary Table S2. No language restrictions were imposed.

Two independent researchers (HL and HJ) collected the studies using Endnote X9 (Clarivate Analytics, Philadelphia), reviewed the title, abstract, and entire original text, and included relevant studies according to the inclusion and exclusion criteria. In the event of disagreement between the two reviewers regarding eligibility, they reviewed the article together to reach a consensus. If necessary, a third reviewer (N-YH) functioned as an arbitrator. After the entire text of the selected studies was confirmed, study information, including the first author, publication year, country, diagnostic criteria, interventions, treatment duration, results, and adverse events, was extracted and tabulated using Microsoft Excel (2019).

Two reviewers (HL and HJ) independently assessed the risk of bias using the Cochrane Collaboration tool (Higgins et al., 2022). Seven items were evaluated: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. Each item was classified into three categories of risk of bias: “high,” “unclear,” and “low.” If the opinions of the two researchers differed, they attempted to reach a consensus through discussion by consulting a third reviewer (N-YH).

Statistical analyses were performed using the Review Manager program (RevMan, computer program, Version 5.4.1, The Cochrane Collaboration, 2020). Quantitative data from two or more studies was integrated using a synthesis for the same outcome measures; if it was not appropriate, a qualitative description of each study’s data was presented in a tabulation format. To measure the effect size, the risk ratio (RR) and 95% confidence interval (CI) were used for dichotomous data, such as TCE and incidence of AEs, whereas the mean difference (MD) and 95% CI were evaluated for continuous data, such as DSS. The I ² statistic was used to assess the heterogeneity of the included studies. The random-effects model was used for the meta-analysis when it was judged that significant heterogeneity existed (I ² value of greater than 50%); otherwise, the fixed-effects model was used. For each outcome measure, forest plots were presented to display the estimated effect sizes from the included studies.

Subgroup analysis was performed with a sufficient number of studies according to the type of intervention and treatment duration to adjust for heterogeneity. Sensitivity analyses were conducted to determine the robustness of the outcomes and other sources of heterogeneity.

A funnel plot was used to evaluate publication bias if more than ten studies were included in the analysis.

The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) tool was used to evaluate the quality of cumulative evidence in this review (Goldet and Howick, 2013). By assessing the factors such as the risk of bias, inconsistency, and indirectness, the certainty level of evidence was judged using four categories of “very low,” “low,” “moderate,” and “high.”

As of 31 January 2023, 396 articles were retrieved from 11 databases, and 13 articles were identified through manual searches. After excluding 19 overlapping articles, 390 were selected. After screening the titles and abstracts of the included articles, 321 were excluded because they did not meet the inclusion criteria. Among the 66 full texts retrieved, 3 studies were duplicate records, 25 were not involved in the relevant population, 27 did not use an appropriate intervention, and 5 were non-clinical trials. Finally, six studies (Li, 2006; Li et al., 2006; Xu, 2007; Wang, 2014; Wu, 2014; Liu and Xie, 2017) including 707 subjects were included in this systematic review (Figure 1) (Page et al., 2021).

All six included articles involving 707 patients were published in China between 2006 and 2017 (Table 2). For the study design, five (Li, 2006; Xu, 2007; Wang, 2014; Wu, 2014; Liu and Xie, 2017) were two-arm, parallel-design RCTs, and one (Li et al., 2006) was a three-arm trial comparing XYS, WM, and combination therapy. Among the five trials, three (Xu, 2007; Wang, 2014; Wu, 2014) compared XYS with WM, and two (Li, 2006; Liu and Xie, 2017) compared combination therapy with WM alone. Regarding the type of XYS, four studies (Xu, 2007; Wang, 2014; Wu, 2014; Liu and Xie, 2017) used XYS, and two (Li, 2006; Li et al., 2006) used DZXYS in the intervention group. Herbal prescriptions and their components in the included RCTs are presented in Table 3. Regarding the type of WM, two studies (Li et al., 2006; Wu, 2014) used Mosapride, and two (Xu, 2007; Wang, 2014) used Domperidone plus enzyme compounds in the control group to compare XYS with WM. In comparing the combination therapy and WM groups, one study (Li, 2006) compared XYS plus Tibolone versus Mosapride plus Tibolone, and the other (Liu and Xie, 2017) compared XYS plus AD versus Mosapride plus AD. The number of patients in the included studies ranged from 80 to 209, with an average of 118, and the trial duration ranged from 2 weeks to 2 months, with an average of 4 weeks. All studies used the XYS decoction as the dosage form. As for the diagnostic criteria for FD, three studies adopted Rome II (Li, 2006; Li et al., 2006; Xu, 2007), and three adopted Rome III (Wang, 2014; Wu, 2014; Liu and Xie, 2017). Five trials used XYS in the treatment of FD according to pattern identification (PI), including Disharmony of liver and spleen systems pattern (Li, 2006; Li et al., 2006; Liu and Xie, 2017), Disharmony of liver and stomach systems pattern (Wang, 2014), and Tangled cold and heat pattern (Xu, 2007).

The risk of bias is presented in Figures 2, 3.

Of the six studies, three mentioned AEs; one study listed only the types of AEs, regardless of the interventions, while two reported detailed information. Li et al. (2006) reported mild headache, dizziness, and insomnia during the study period. Wu (2014) reported that mild diarrhea occurred in the XYS group and dizziness and mildly elevated transaminases occurred in the Mosapride group. Li (2006) reported that increased stool frequency occurred in the DZXYS combined with Tibolone group and mild diarrhea, mild headache and dizziness, and insomnia in the Mosapride combined with Tibolone group. A meta-analysis of two studies showed that there were significantly fewer AEs in the experimental group (XYS and combination therapy) than in the control group (WM alone) (RR = 0.20, 95% CI: 0.07, 0.63, p = 0.006), with low heterogeneity (I ² = 45%, p = 0.18) (Figure 8).

Because each synthesis contained no more than 10 articles, publication bias was not analyzed using a funnel plot in this review.

The certainty of the meta-evidence for the efficacy of XYS for the patients with FD was “high” on AEs; “moderate” on TCE, DSS, and upper abdominal pain; “low” on abdominal distension; and “very low” on belching and early satiety (Table 4). The risk of bias in all included studies was judged serious, as a high-performance bias was observed. The high heterogeneity across studies and the imprecision of the results in meta-analyses have resulted in a downgraded level of evidence.

In this study, a systematic literature review and meta-analysis was performed to evaluate the efficacy and safety of XYS as a monotherapy and as an adjuvant treatment to conventional WM for FD. Six RCTs with 707 patients were included in this study. As critical outcomes, the TCE and DSS results showed a significant improvement in the XYS and combination therapy groups compared to the WM alone group. In particular, for the symptoms of abdominal distension and upper abdominal pain, XYS significantly alleviated the severity of symptoms compared with Mosapride when used alone or in combination with Tibolone. The above two symptoms are common in FD, which suggests that XYS can be used as an alternative therapy for FD patients who show an insufficient response to conventional WM treatment such as PK or prefer traditional medicine. Additionally, fewer adverse events were reported in the XYS group than in the WM group.

To specifically identify the effects of XYS, the subgroup analysis was performed involving the administration of DZXYS. As a result, DZXYS monotherapy and in combination with Tibolone may be an option for perimenopausal individuals with FD whose symptoms are poorly managed by customary treatments and are prone to recurrence.

FD is the most common FGID and is related to abnormal neuromodulation of the brain-gut axis, disharmony in the gut microbiome, and an imbalance of gastrointestinal hormones (Mertz, 2003; Fukui et al., 2018). In traditional Chinese medicine (TCM), XYS is used for dyspeptic symptoms, including stiffness or pain in the upper abdomen, reduced intake, and feelings of malaise or tiredness, which are caused by the pathophysiological pattern of Disharmony of liver and spleen systems. In TCM, XYS relieves these symptoms by circulating qi. In particular, the indication for XYS is closely related to Liver qi stagnation, expressed as emotional instability that can occur in conditions such as depressive disorders and postmenopausal syndrome (Lee and Jeong, 2017). In a meta-analysis of 26 studies with 1,837 patients with depression, combination therapy of XYS and AD significantly improved depressive symptoms compared to the AD alone group. These results suggest that XYS is clinically effective in relieving psychological symptoms such as anxiety and depression (Zhang et al., 2012). Recently, the antidepressant effect of XYS has been shown to be associated with brain-gut peptides (Liu et al., 2017), regulation of the gut microbiome (Zhu et al., 2019), inhibition of the hypothalamic-pituitary-adrenal axis (Zhu et al., 2014; Yan et al., 2018), and intracerebral interactions responsible for emotion and perception (Malagelada, 2020). The chemical compositions and pharmacological effects of the single components of XYS are presented in Supplementary Table S3. Among the six studies in this review, five performed PI before administration, which is defined as the process of determining the cause and nature of a patient’s disease in TCM (World Health Organization and Regional Office for the Western Pacific, 2007). Overall, Disharmony of liver and spleen/stomach systems pattern, which is related to the symptoms of indigestion accompanied by psychological depression, was the most frequently used pattern in the included studies. This suggests that XYS is recommended for FD patients with anxiety or depression by improving mental health.

Unlike previous reviews (Qin et al., 2009; Tang et al., 2019), this review analyzed the effect of XYS combined with WM and included various conventional WM, including PK, Tibolone, and AD. We only included studies that described herbal prescriptions in detail on administration information and components for clarity. In addition, the data search was performed comprehensively not only in China but also in global and Asian countries, including Japan and Korea, to reflect the results of clinical studies in countries with a high frequency of XYS administration for FD. Finally, unlike prior meta-analyses, our research demonstrated that DZXYS was not inferior to WM in the treatment of FD in perimenopausal females.

This study has several limitations. First, there was a regional bias because all the studies in this review were based in China. Second, the quality of the included studies was low. The risk of bias in most of the studies were rated as “moderate” or “high” due to insufficient information on randomization process, allocation concealment, and methods of blinding outcome assessors. In addition, there were no previously published study protocols to assess other biases. In particular, double-blinding was not conducted in all of the included studies. When both herbal decoction and WM are used as interventions, it is difficult to perform blinding due to their differences in external characteristics. Therefore, in future studies, it is necessary to plan and adhere to a strict blinding protocol, such as using a double-placebo, and describe the method in detail. Third, during the selecting process, a significant number of studies that did not meet the intervention or diagnostic criteria were excluded from this review; thereafter, the number of included studies was insufficient to resolve the heterogeneities of intervention and treatment duration. Fourth, some meta-analyses were not available because studies that reported gastrointestinal hormones and gastric emptying rate or used validated questionnaires for dyspeptic or psychological symptoms were rare. There is no specific marker for FD and the patient complaints and symptoms are often ambiguous, making it difficult to quantify the improvement in dyspeptic symptoms. To resolve this ambiguity, various scales have been developed to measure the severity of FD symptoms. In this review, a meta-analysis of DSS showed that XYS significantly improved discomfort in the upper abdomen compared with WM. However, other variables could not be analyzed because no studies used standard questionnaires for FD, such as the Nepean Dyspepsia Index (Talley et al., 1999), the FD-related quality of life questionnaire (Lee et al., 2006), or the gastrointestinal symptom score (Adam et al., 2005). Further research using various validated outcome measures is required to compensate for these limitations. In addition, as for psychological symptoms, the results of analyzing HAM-A and HAM-D between the intervention and control groups were not reported due to lack of research. Fifth, the treatment duration and follow-up periods for assessing recurrence rates differed between studies, making it difficult to evaluate the persistent effect of XYS. Finally, the review found that quality control and chemical profile were not reported in any of the included studies, which has led to controversial conclusions. To address this issue, future clinical trials on botanical drug extracts should provide detailed information on the composition, extraction process, and drug extraction ratio of the study material. Additionally, presenting the basic pharmaceutical parameters of the extracts would allow for both qualitative and quantitative comparison.

Compared to conventional WM, widely used to manage FD symptoms, both XYS and combination therapy showed significant efficacy and fewer adverse events. Thus, XYS may be administered to patients who do not respond to WM alone or who show adverse effects. Considering the low quality of the included studies, rigorous large-scale RCTs are required to confirm our findings. To strengthen the evidence for the effect of XYS in patients with FD, well-designed RCTs should be performed evaluating various outcome measures related to dyspepsia. Furthermore, future studies should assess the effectiveness of XYS with long-term follow-up.