We searched 11 databases: Medline through PubMed, Embase through Elsevier, Web of Science, Cochrane Central Register of Controlled Trials, SCOPUS, Allied and Complementary Medicine through EBSCO, PsycARTICLES through EBSCO, Korea Citation Index, Oriental Medicine Advanced Searching Integrated System, Citation Information by NII, and China National Knowledge Infrastructure. Furthermore, bibliographies of the retrieved articles and pertinent systematic reviews were manually searched. Two independent researchers conducted a literature review on July 22, 2024.

The search strategy incorporated different synonyms and related medical subject headings. The search strategy incorporated different synonyms and related medical subject headings (Supplementary Material 1).

Two independent researchers conducted individual searches and compared the results to ensure that no studies were missed. This step was performed using the reference management program EndNote 20.2.1 (Clarivate Analytics, Philadelphia, PA, USA). In cases of disagreement, the two researchers reached a consensus; if disagreement could not be addressed, a third researcher was consulted to make the final decision regarding inclusion.

Two independent researchers used Excel 2016 (Microsoft, Redmond, WA, USA) to extract and crosscheck the data. In cases where data were missing, the corresponding author was contacted via email to obtain data. In cases of disagreement, a third researcher was consulted to reach a consensus and make the final decision.

Two independent researchers evaluated the detailed items using Cochrane’s Risk of Bias (RoB) tool.²⁸⁾ The following evaluation domains were considered: the randomization process, deviations from the intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. The risk of bias for each domain was graded as follows: high risk, low risk, or some concerns. Results of the assessment were summarized using a risk of bias graph and a risk of bias summary figure. Any disagreement between the two authors was resolved through discussion with the third experienced author of the review (SHK). In cases of disagreement, a third researcher was consulted to reach a consensus and make the final decision.

Relative risk (RR) was used for dichotomous variables, and the mean difference (MD) or standardized mean difference (SMD) with a 95% confidence interval (CI) was used for continuous variables. For HAMA, MD was used. Because various HAMD scales (17 or 24) were used, which are both major measures, SMD was used. Owing to the heterogeneity in the participants’ demographic characteristics, severity of depression and anxiety, prescriptions used, control groups, and treatment duration in the included studies, clinical heterogeneity was assumed to be significant. Therefore, regardless of the I² value, a random-effects model was used for the post-analysis. However, if the number of included studies was very small, a fixed-effects model was employed for the post-analysis. The Review Manager Software version 5.4 (Copenhagen, The Nordic Cochrane Centre, the Cochrane Collaboration, 2020) was used for summary and synthesis of the effects of studies with the same intervention and control groups. Subgroup analysis was conducted based on the treatment duration (4, 6, 8, and 12 weeks) and the types of medications used in the control group: SSRI and flupentixol/melitracen. The treatment duration was classified into three phases: early response (1 to 4 weeks), acute phase treatment (6 to 12 weeks), and follow-up response (16 to 24 weeks).²⁹⁾ We conducted a sensitivity analysis by excluding studies evaluated as having a high RoB according to selection bias and those that did not overlap with the CI of the overall effect.

We evaluated the evidence level of the literature included in the meta-analysis using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) methodology.³⁰⁾

A total of 12,008 articles were identified through the literature search. After reviewing titles and abstracts, we conducted an initial screening to exclude duplicates (n=922) and articles that did not meet the selection criteria (n=10,784). For the 302 articles selected in the first screening, we obtained full texts and conducted a second screening. During the second screening, we excluded 285 articles that were not RCTs (n=59), did not focus on MDD (n=141), had inappropriate interventions in the treatment (n=66) or control groups (n=5), had inappropriate outcomes (n=2), or were duplicate articles (n=12). Finally, 17 RCTs were included in this study (Figure 1).^31–47)

The characteristics of the included studies are shown in Table 1. All RCTs were conducted in China. Sample sizes ranged from 40 to 132, with participants’ mean age ranging from 69.06 ± 4.28 to 82.8 ± 6.7 years in the intervention group and 69.47 ± 1.26 to 85.1 ± 14.8 years in the control group. Treatment duration ranged from 15 days to 6 months, averaging 6.95 weeks. No follow-ups were performed. Eight studies used TEAHM as the sole treatment intervention,^31–38) while nine used TEAHM with antidepressants.^39–47) Twelve studies used CCMD-3 as the diagnostic criterion for MDD, ^{31–35,38,39,41,42,44,46,47)} three used ICD-10,^35,43,45) and one used the DSM-IV criteria.³⁶⁾ One study used CCMD-3 and ICD-10.³⁵⁾ Sixteen studies used HAMA to assess anxiety symptoms^{31–34,36–47)} and one used the Zung SAS.³⁵⁾ Six studies conducted pattern identification.^{35,36,39,41,45,47)} In most studies, participants had anxiety assessment of a HAMA score ≥14 or an SAS score >53. The baseline severity of anxiety symptoms in HAMA ranged from 18.5 ± 3.5 to 28.41 ± 4.50. This indicates that most participants presented with moderate anxiety at baseline. HAMD, SDS, and TER were used as evaluation tools to assess depressive symptoms. TCM syndrome scores were evaluated in three studies, and the incidence of adverse effects was used to evaluate side effects. Nine herbal prescriptions were administered to patients in each treatment group. Among them, the most commonly used prescription was the Shugan Jieyu capsule/decoction, which was used in six studies.^{37,38,42,43,45,46)} This was followed by Oryeong-san (Wuling capsule), used in three studies,^31,34,44) and Gamisoyo-san (Jiawei Xiaoyao capsule), used in two studies.^36,40) The most frequently used single herbal medicine was Sclerotium of Poria cocos WOLF, which was used six times. The single herbal medicines used as ingredients in nine different TEAHM prescriptions were listed and analyzed based on their frequency of use (Supplementary Materials 2 and 3). Herbs such as Bark of Albizzia julibrissin DURAZZ., Root of Angelica gigas NAKAI, and Root of Glycyrrhiza uralensis FISCH. were each referenced three times. Five studies used paroxetine as the conventional treatment.^39,41–44) This was followed by two studies that used fluoxetine,^33,47) sertraline,^36,37) and one study that used flupentixol/melitracen (Deanxit).⁴⁵⁾ The treatment duration ranged from 4 to 12 weeks.

The methodological quality of the 17 included studies, analyzed using the Cochrane RoB, is summarized in Figure 2. Seven studies that used randomization were evaluated as “low” for selection bias.^{33,36–39,41,45)} The remaining 10 studies that did not mention randomization or random allocation were evaluated as “high” for selection bias.^{31,32,34,35,40,42–44,46,47)} One study that used the envelope method to conceal the allocation area was evaluated as “low” for selection bias,³⁶⁾ and one study that used the open envelope method was evaluated as “high” for selection bias.³⁵⁾ The remaining 15 studies that did not indicate if the allocation was concealed were evaluated as “unclear” for selection bias.^{31–34,37–47)} One study that used a double anonymization method to anonymize the participants was evaluated as “low” for selection bias,³⁶⁾ and the study that used a single anonymized procedure was evaluated as “low” for selection bias.³⁵⁾ The remaining 15 studies that did not indicate if the participants were anonymized were evaluated as “high” for selection bias.^{31–34, 37–47)} One study with anonymization of outcome assessment was evaluated as “low” for selection bias,³⁶⁾ and the remaining 16 studies that did not mention whether the outcome assessment was anonymized were evaluated as “high” for selection bias.^{31–34,37–47)} Regarding attrition bias, 14 studies with no dropouts or missing data were evaluated as “low” for selection bias.^{31,32,34,35,37–42,44–47)} Three studies that did not provide information on dropouts were evaluated as having a high risk of selection bias.^33,36,43) Regarding reporting bias, 3 studies were evaluated as “low” for selection bias, ^35,36,41) and 14 studies where the results were reported according to the protocol were evaluated as “unclear” for selection bias.^{31–34,37–40,42–47)} In the other bias areas, studies with insufficient information were evaluated as “unclear.”^31–47)

A meta-analysis was performed for the primary outcome (HAMA) and secondary outcomes (HAMD and TER) for depression. A subgroup analysis was conducted based on the treatment duration and medication type used in the control group. Studies with fewer than one publication or unanalyzable results for the SDS,³⁵⁾ SAS,³⁵⁾ and TESS^{31,32,38,42,45,47)} scales were not included in the meta-analysis.

The overall quality of the evidence varied from low to moderate. Between TEAHM alone and antidepressants, the quality of evidence was assessed as low. Moreover, between TEAHM combination therapy with antidepressants and antidepressants alone, the quality of evidence ranged from low to moderate. The lack of participant anonymization, inconsistency, and imprecision were the reasons for this downgrade (Table 2).

This review and meta-analysis provides evidence regarding the effectiveness and safety of TEAHM in treating patients with MDD and anxiety. Seventeen studies were included after screening (Eight studies using TEAHM alone and nine studies using TEAHM and antidepressants).^31–47) The main results are as follows. (1) Compared with antidepressants alone, TEAHM monotherapy significantly improved HAMA scores. This combination also showed a more significant improvement in HAMA scores compared to those with antidepressants alone. (2) Compared with antidepressants alone, TEAHM monotherapy or combination therapy significantly improved HAMD and TER scores. (3) In the subgroup analysis of periods and antidepressants, the anxiolytic effects of TEAHM monotherapy and combination therapy were superior to those of SSRI, DNRI, and TCA in patients with MDD. (4) TEAHM alone had a lower risk of adverse events, and combination therapy exhibited a similar risk. However, no severe adverse events were reported. (5) The treatment duration ranged from 4 to 12 weeks. The most commonly used prescription was the Shugan Jieyu capsule, and the most frequently used single herbal medicine was Sclerotium of Poria cocos WOLF. (6) The overall RoB across all studies was high due to the bias in allocation concealment, participant and outcome assessor anonymization, and reporting bias domain. The quality of the evidence obtained using GRADE varied from very low to moderate.

Previous studies have shown that compared with control medication, Chinese herbal medicine (CHM) significantly improves HAMA scores, with fewer adverse events in the management of generalized anxiety disorders, a common anxiety disorder.⁴⁸⁾ Single herbs, such as Ginkgo biloba and kava, also significantly improve HAMA scores.¹⁹⁾ CHM monotherapy and CHM with conventional pharmacotherapy have also shown significantly better HAMA scores in patients with post-stroke anxiety and a lower incidence of adverse events.⁴⁹⁾ In addition, Soyo-san (Xiaoyao San), a well-known CHM, used alone or in combination with anxiolytic agents, showed greater effectiveness and safety than did anxiolytic treatment for anxiety disorders.⁵⁰⁾ Fewer adverse reactions were observed with Soyo-san monotherapy as well as Soyo-san combination therapy with anxiolytic treatment than with anxiolytic treatment alone.⁵⁰⁾ Our findings showed that compared with antidepressant therapy alone, CHM combination therapy did not improve adverse effects; however, adjunctive therapy with Soyo-san improved adverse effects.⁵⁰⁾ The patients in these reviews were not patients with MDD but those with anxiety disorders; however, the findings of our review align with the results of these studies. TEAHM significantly improved HAMA scores in patients with MDD and anxiety, with fewer adverse events.

Furthermore, TEAHM improved anxiety and depression in patients with MDD. A recent study reported that CHM showed effects similar to those of antidepressants in terms of improving HAMD scores and TER in patients with depression.¹⁸⁾ In addition, CHM combined with antidepressants improved HAMD scores and led to a higher TER than did antidepressant monotherapy.¹⁸⁾

Moreover, patients who underwent CHM monotherapy and combination therapy showed fewer adverse events than did antidepressant monotherapy.¹⁸⁾ These results are consistent with the findings of the present review. Therefore, TEAHM may be a potential alternative or complementary treatment option to improve depression and anxiety in patients with MDD without severe adverse events.

Shugan Jieyu capsule (SJC) was the most commonly used prescription among the included studies. Clinical studies have shown that using SJC alone or in combination with conventional pharmacotherapy can significantly improve psychiatric symptoms in patients with post-stroke depression.⁵¹⁾ Based on a recent network meta-analysis for HAMD score reduction response rate, SJC plus SSRI was one of the most effective prescriptions for post-stroke depression.⁵²⁾ This review also showed that SJC was safer than SSRI in terms of adverse events.⁵²⁾ SJC improves various symptoms, including psychic anxiety, low mood, automatic nerve dysfunction, and somatic anxiety.⁵³⁾ Shugan Jieyu capsules also treat potential diseases such as primary and concomitant comorbid anxiety with depression, physical diseases, and postpartum depression.⁵³⁾ Compared with conventional pharmacotherapy, Shugan Jieyu capsules significantly improved sleep quality and reduced adverse reactions in patients with insomnia.⁵⁴⁾ These results are consistent with those of the present review.

First, this review provides clinical evidence supporting the use of TEAHM as a complementary intervention for managing anxiety symptoms in patients with MDD. Conventional pharmacotherapy faces challenges such as adverse effects, low treatment adherence, and suboptimal treatment responses.¹⁵⁾ Although Cognitive Behavioral Therapy is widely recommended as a non-pharmacological alternative, its availability remains limited.^14,17) For patients seeking complementary or alternative approaches, the findings of this review may assist clinicians in making evidence-based decisions. Second, our findings showed that combining antidepressants with TEAHM resulted in significantly greater improvements in anxiety symptoms compared to antidepressant monotherapy. This suggests that TEAHM may serve as a valuable adjunctive option, especially for patients who exhibit partial response or intolerance to conventional antidepressants. Third, this study offers a detailed summary of the specific TEAHM prescriptions and the frequency of use of individual herbal components across trials. Such data can inform clinical practice by identifying the most frequently applied herbal combinations, which may serve as a foundation for the development of practical treatment protocols in routine care. Fourth, some of the included studies reported pattern identification data—a core diagnostic concept in East Asian traditional medicine that classifies patients based on detailed symptom patterns.⁵⁵⁾ This can aid clinicians in stratifying patients with MDD and anxiety symptoms, enabling more personalized and precise treatment approaches. Fifth, an important clinical implication of this review is its insight into the safety profile of TEAHM when used in combination with conventional pharmacotherapy. While standard options like benzodiazepines are associated with risks such as dependence and cognitive impairment,¹⁴⁾ the included studies reported a relatively favorable safety profile for TEAHM. This suggests that TEAHM may be particularly useful for the long-term management of anxiety symptoms in MDD, where minimizing side effects and reducing dependency risks are critical considerations.

This review was subject to certain limitations, which should be considered when interpreting the results of this review. First, MDD is highly heterogeneous, and the DSM-5 introduced several specifiers to subdivide this broad and complex diagnosis into more precise categories. Although there is a description of anxious depression as a specifier of MDD in the DSM-5, there are still no clear categorical criteria regarding MDD with anxiety.¹³⁾ Additionally, except for two studies, none of the included studies reported using the criteria regarding anxious depression as the specifier of MDD in the DSM-5 or specific points on HAMA as inclusion criteria. For meta-analyses of particular medical conditions with comorbidity, it is critical to use relevant assessment tool for comorbidity to justify study. Although most included studies use HAMA as inclusion criteria for MDD with anxiety, no clear criteria for anxious MDD may result in highly heterogeneity across the participant population. Furthermore, distinguishing Anxious Depression from an Anxiety Disorder is challenging.⁵⁶⁾ These issues may reduce the robustness and generalizability of our findings. Second, moreover, the primary outcome of the included studies was depression rather than anxiety. Notably, most studies have aimed to evaluate the effectiveness of TEHAM in treating depression rather than anxiety alone in patients with MDD. Third, the studies included were evaluated as “low” regarding methodological quality, particularly owing to selection bias due to improper randomization and allocation and performance bias due to the lack of placebo-controlled trials. This limitation indicates that this review may have overestimated the effectiveness of TEAHM. Moreover, the quality of evidence was not high according to the GRADE evaluation, indicating that our results have low reliability. Therefore, the evidence presented in this review should be interpreted cautiously because it is subject to changes following the future findings of high-quality studies. Fourth, factors such as patient sex and age, severity of depression, duration of depressive episodes, duration of comorbid anxiety, pattern identification, prescription variations, treatment duration, and antidepressant types were heterogeneous across the included studies. Therefore, this limitation could significantly affect the results of the meta-analysis, regardless of statistical heterogeneity. Except for the duration of treatment and type of antidepressant used in the included studies, a subgroup analysis based on the above variables was not feasible due to the limited number of studies. Fifth, subgroup analysis for different TEAHM prescriptions was not conducted; therefore, differences in effectiveness according to different prescriptions could not be confirmed.

Our suggestions for future research are as follows. First, future research needs to use the specifier of MDD in the DSM-5 as the criteria for anxious depression and report clear specific points of the HAMA used as inclusion criteria. Notably, more studies on the characteristics and specific definitions of anxious depression as a categorical diagnostic criterion are required.⁵⁷⁾ Second, large multicenter, double-anonymized, placebo-controlled trials are essential to evaluate the efficacy and safety of herbal medicines and to avoid potential placebo effects. A longer treatment duration and follow-up should be conducted to prove the sustainability of TEAHM’s effectiveness and safety. Standard guidelines for future clinical trials should be developed to reduce clinical heterogeneity. A study by Su et al. provides a useful model for designing and conducting robust RCTs in the future.³⁶⁾ A prospective cohort study of MDD would also help address such heterogeneity-related issues.⁵⁸⁾ Third, clinical trials in different countries and cultures should be conducted, since all the included studies were performed in China. International cooperative research on this topic, such as the China–India Mental Health Alliance, should be strengthened between East Asia and other countries using traditional, complementary, and alternative medicines.⁵⁹⁾ Fourth, if more studies are included in a future meta-analysis, a subgroup analysis according to different TEAHM prescriptions should be conducted to confirm the effectiveness of each prescription as evidence for prescription selection. Finally, further research is required regarding the adverse effects of herbal-drug interactions between TEAHM and antidepressants.