This study primarily utilized the China National Knowledge Infrastructure (CNKI) and the Web of Science (WOS) Core Collection as data sources to retrieve literature related to EA treatment for depression.

The CNKI database retrieval strategy was: SU = (“抑郁” + “郁证” + “抑郁综合征” + “抑郁症” + “神经官能性抑郁症” + “忧郁症” + “单相抑郁症” + “抑郁性神经症” + “内源性抑郁症” + “内因性抑郁症” + “抑郁综合征” + “CUMS” + “慢性不可预知刺激”) * (“电针” + “电针疗法” + “脉冲电针疗法” + “电针疗法” + “电热针疗法” + “福尔电针疗法” + “穴位电刺激”). The WOS Core Collection retrieval strategy was: TS = (depression OR depressive disorder OR Chronic unpredictable mild stress OR CUMS OR Poststroke depression OR postnatal depression OR Climacteric depression OR menopause depression OR adolescent depression) AND TS = (electroacupuncture OR Electroacupuncture EA OR Electroacupuncture therapy). The retrieval period spanned from January 2003 to September 2023, based on the publication date.

2.3.1. Inclusion Criteria

Publications with complete bibliographic records, including essential information such as publication date, author(s), title, abstract, keywords, and research affiliations.

2.3.2. Exclusion Criteria

1) Duplicate publications; 2) publications whose research topics were irrelevant to the present study; 3) publications not in the form of journal articles, such as conference proceedings, reports, scientific and technological achievements, technical standards, and patents.

Two researchers independently reviewed the literature and cross-checked their selections. Chinese and English publications were included according to the inclusion and exclusion criteria. Any disagreements were resolved through discussion with a third researcher. The screened Chinese publications were exported as text files in RefWorks format, named “download_***” for subsequent use. The screened English publications were exported in the “Full Record and Cited References” format within “Plain Text File” for subsequent use. All retrieval, screening, and export procedures were completed within one day.

Data were imported into Cite Space 6.2.R4. The time span was set to 2003-2023, with “Years Per Slice” set to 1. Node types were set to author, institution, and keyword, respectively. For institutional co-occurrence, the g-index (k) was set to 10, Top N was set to 4, and other parameters remained at system defaults to generate maps. For keyword co-occurrence, the g-index (k) was set to 10, with other parameters remaining at system defaults to generate keyword co-occurrence, clustering, and burstness maps.

Data were imported into VOSviewer 1.6.18. The full counting method was selected, and the minimum number of documents was set to 2. “Co-authorship” was selected under “Type of analysis” and “Documents” was selected under “Weights” to construct author collaboration network maps and author collaboration timeline network maps weighted by publication count for both CNKI and WOS databases.

The exported .txt files were compressed into .zip format and imported into the Bibliometrix bibliometric analysis web interface running in the R language environment. The “Country Collaboration” function in the sidebar was selected to analyze the imported literature. The bibliometric parameters including g-index (k = 10), Top N = 4 (CiteSpace), and minimum documents = 2 (VOSviewer) were selected based on standard bibliometric mapping practices recommended by Chen’s classic CiteSpace methodological framework and the official VOSviewer tutorial for medical literature analysis. The g-index k = 10 balances the visibility of core collaborative nodes without overloading networks with low-impact institutions/authors; Top N = 4 limits institution co-occurrence maps to high-frequency institutions to avoid visual clutter. The threshold of minimum documents = 2 excludes one-off isolated publications, which are widely adopted in acupuncture bibliometric studies to filter noise. A simple sensitivity test was performed: adjusting k from 8 to 12 and minimum documents from 1 to 3 did not alter the core clustering structure of author/institution networks, confirming the robustness of selected parameters ([5]).

A preliminary search of the selected databases was conducted according to the search strategies for publications from January 1, 2003, to September 25, 2023. A total of 1469 publications were retrieved from the Chinese database and 455 from the English database, yielding 1924 publications. After screening, 881 valid Chinese publications and 148 valid English publications were obtained. The literature screening flowchart and results are presented in Figure 1.

Figure 1. Literature screening flowchart.

Analysis of the annual publication volume over the past 20 years revealed that the average annual number of Chinese publications related to EA treatment for depression was approximately 42. From 2003 to 2014, the overall trend showed a fluctuating upward pattern, with peak publication volumes of approximately 60 articles in 2014, 2016, and 2017. Thereafter, the annual publication volume remained generally stable at around 50 articles, as detailed in Figure 2. The average annual number of English publications related to EA treatment for depression was approximately 7, exhibiting an overall upward trend. The largest increase occurred between 2018 and 2019, with the publication volume peaking at 21 articles in 2021. Combined analysis of Chinese and English publication volumes indicates that over the past 20 years, the number of publications on EA treatment for depression has fluctuated, exhibiting a trend of Chinese publications stabilizing at a high level and English publications increasing year by year. Although English publications demonstrate a general upward trajectory across 2003-2023, minor declines occurred during 2016-2017 and 2021-2022, which may be attributed to periodic fluctuations in research output and delayed publication cycles of clinical trials.

The included English literature was analyzed using the Bibliometrix package in the R environment. As shown in Figure 3, research in the field of EA treatment for depression is currently predominantly conducted in China. Although China has collaborations with other countries such as the United States and the United Kingdom, the collaboration networks formed are relatively sparse and the collaborative relationships are not close.

Figure 2. Annual publication trend chart of literature on electroacupuncture therapy for depression over the past 20 years.

Figure 3. National cooperation relationship network diagram.

Analysis of the included literature using VOSviewer revealed that the Chinese publications involved 1779 authors, of whom 1305 (73.36%) published only one article. The English publications involved 862 authors, of whom 719 (83.41%) published only one article. The proportion of low-productivity authors exceeded the 60% threshold defined by Lotka’s Law ([24]). The number of core authors was calculated using Price’s Law formula: M = 0.749 × (Nmax)1/2, where M represents the minimum number of publications and Nmax represents the publication count of the most prolific author in the dataset. Authors with publication counts exceeding M were defined as “core authors” ([29]; [21]). In CNKI, Mmax = 48, N ≈ 5; 72 authors published more than five articles, accounting for 4.05%. In WOS, Mmax = 12, N ≈ 3; 59 authors published three or more articles, accounting for 6.84%. These findings indicate that the research authors in both CNKI and WOS databases are numerous and dispersed, forming a core author-driven team development model. Researchers in this field may strengthen collaboration in the future to conduct more in-depth and sustained studies.

3.4.1. Analysis by the Lead Author

Table 1 and Table 2 list the top 10 authors by publication volume in EA treatment for depression research in the CNKI and WOS databases, respectively. The publication volume of scholars in CNKI was significantly higher than that of scholars in WOS. In CNKI, the top three authors by publication volume were Tu Ya (48 articles), Li Zhigang (28 articles), and Ren Lu (17 articles), with their combined publications accounting for 10.5% of the total. In WOS, the top three authors were Huang, Yong (12 articles), Lao, Lixing (12 articles), and Xu, Shifen (11 articles), with their combined publications accounting for 23.6% of the total. These core authors have made outstanding contributions to EA research on depression.

Table 1. Publication count scale for authors of Chinese literature on electroacupuncture therapy for depression.

Table 2. Scale for the number of publications by authors of English literature on electroacupuncture treatment for depression.

Furthermore, the core authors in this field are all of Chinese nationality. The majority of the top 10 authors in CNKI are based in northern China, whereas the majority of the top 10 authors in WOS are based in southern China. This suggests that geographical factors may exert an influence, resulting in limited collaboration between scholars from northern and southern China.

3.4.2. Author Collaboration Analysis

VOSviewer was used to generate author collaboration network maps (Figure 4) and author collaboration timeline network maps (Figure 5) weighted by publication count for both Chinese and English literature. The CNKI map displayed 167 nodes and 897 links, while the WOS map displayed 93 nodes and 464 links.

1) Analysis of Author Collaboration Teams

In Figure 4, different colors represent different team clusters. As illustrated, the author collaboration network map for CNKI contained eight clusters, exhibiting a state of “partial concentration with overall dispersion.” The largest team was the collaboration team centered around Tu Ya, followed by the team centered around Li Zhigang. Research between these two major core teams was relatively independent, and the scale of collaboration was also relatively small. The author collaboration map for WOS contained six clusters, with four major core research teams initially formed: the collaboration team centered around Huang, Yong; the team centered around Lao, Lixing; the team centered around Tu, Ya; and the team centered around Yu, Jin. There was a certain degree of collaboration among the major teams, with relatively close connections. In summary, compared with the WOS literature, the author collaboration network in CNKI was more scattered, with academic mentorship relationships (“Xueyuan”) playing a more dominant role than geographical relationships (“Diyuan”).

Figure 4. Collaborative author map of Chinese and English literature on electroacupuncture therapy for depression. (A) Collaboration network of authors in Chinese literature; (B) Collaboration network of authors in English literature.

2) Analysis of Author Publication Timeline

The author publication timeline can reveal emerging collaborative teams, facilitating scholars in better identifying hotspots and potential collaborators in the corresponding field ([26]). Based on the author collaboration network map (Figure 4), selecting the Overlay Visualization option generated the author publication timeline map (Figure 5).

Figure 5. Network diagram illustrating the evolution of publication dates for authors in Chinese and English literature. (A) Network diagram illustrating the evolution of publication timelines for authors in Chinese literature; (B) Network diagram illustrating the evolution of publication timelines for authors in English literature.

In terms of publication timeline, within the two major core collaboration teams in CNKI (Tu Ya’s team and Li Zhigang’s team), scholars with higher publication counts generally made their first publications before 2018. In contrast, Fang Jiliang’s team primarily began publishing after 2020, representing a newly formed academic exchange team in recent years. In WOS, scholars with higher publication counts generally made their first publications before 2018.

For institutional data, only the primary institutional affiliation was included; specific hospital departments and school colleges were removed. Institutions with multiple names were standardized to their most recent names for statistical purposes. After the above reprocessing of the screened data, institutional collaboration network co-occurrence maps for Chinese and English publications were generated separately (Figure 6).

As shown in Figure 6(A), the institutional collaboration network co-occurrence map for Chinese publications contained 290 nodes and 149 links, forming seven major clusters centered around Beijing University of Chinese Medicine, Guangzhou University of Chinese Medicine, Nanjing University of Chinese Medicine, China Academy of Chinese Medical Sciences, Heilongjiang University of Chinese Medicine, Chengdu University of Chinese Medicine, and Shanghai University of Traditional Chinese Medicine. Among the high-productivity institutions, Beijing University of Chinese Medicine had the highest publication volume (128 articles). However, the betweenness centrality values of the publishing institutions were generally low. The highest betweenness centrality value was 0.01 (<0.1) for Beijing University of Chinese Medicine, and the network density was 0.0036. This indicates that although numerous institutions conduct research in the field of EA treatment for depression, inter-institutional collaboration is limited, and closer cooperation needs to be developed.

Figure 6. Collaborative network of Chinese and English literature institutions for electroacupuncture treatment of depression. (A) Collaborative network map of Chinese literature institutions; (B) Collaborative network map of English literature institutions.

As shown in Figure 6(B), the institutional collaboration network co-occurrence map for English publications contained 77 nodes and 121 links, forming major clusters centered around the University of Hong Kong, Beijing University of Chinese Medicine, and Shanghai University of Traditional Chinese Medicine. Among the high-productivity institutions, the University of Hong Kong had the highest betweenness centrality value of 0.35 (>0.1), and the network density was 0.0405, indicating relatively close collaboration among institutions. The institution with the highest publication volume was Shanghai University of Traditional Chinese Medicine (25 articles), which was relatively lower compared with the publication volumes of institutions in the Chinese literature, suggesting that further research is warranted.

3.6.1. Keyword Co-Occurrence

Keywords typically represent the core content of articles. Statistical analysis of keyword frequency can reveal research hotspots within a specific field ([3]). Software parameters were set according to the methods section. After preliminary visualization, synonyms were merged, such as “印堂” (Yintang) into “印堂穴” (Yintang acupoint) and “behavior” into “behaviors.”

The keyword co-occurrence map generated from the included Chinese literature is shown in Figure 7(A), containing 279 nodes and 873 links. Table 3 lists the top 20 keywords. Combining the map and table, in addition to the search terms, keywords such as rat, hippocampus, chronic stress, clinical research, and fluoxetine appeared with high frequency.

The keyword co-occurrence map generated from the included English literature is shown in Figure 7(B), containing 141 nodes and 469 links. Combining Figure 7(B) and Table 3, in addition to the search terms, keywords such as brain, behaviors, acupuncture point, and randomized controlled trial appeared with high frequency. This indicates that the current focus of the English literature lies in clinical trials and depression-related behaviors.

Figure 7. Co-occurrence map of key terms in Chinese and English literature on electroacupuncture treatment for depression. (A) Co-occurrence map of keywords in Chinese literature; (B) Co-occurrence map of keywords in English literature.

Table 3. Frequency distribution of the top 20 high-frequency keywords related to electroacupuncture treatment for depression.

3.6.2. Keyword Cluster Analysis

Based on the keyword co-occurrence network maps, the LLR (Log-Likelihood Ratio) algorithm was employed to conduct cluster analysis of keywords (Figure 8), aiming to illustrate the research hotspots within the field of EA treatment for depression.

As shown in Figure 8(A), the keyword co-occurrence map for Chinese literature yielded a modularity (Q) value of 0.5355 (Q > 0.3 indicates a significant clustering structure) and a mean silhouette (S) value of 0.8463 (S > 0.7 indicates that the clustering is efficient and convincing) ([5]). Cluster #0 pertained to disease types; Clusters #1, #4, and #5 were related to different acupuncture treatment modalities; Cluster #2 involved treatment methods frequently combined with EA therapy; Cluster #3 represented experimental animals commonly used in research; and Cluster #6 pertained to acupoints commonly stimulated by EA.

As shown in Figure 8(B), the keyword co-occurrence map for English literature yielded a Q value of 0.5048 (>0.3) and an S value of 0.7906 (>0.7), indicating that the clustering was also reasonable and effective ([5]). Cluster #0 represented commonly used treatment methods for depression; Cluster #1 pertained to disease types; Cluster #2 involved acupoints commonly stimulated by EA; Cluster #3 was related to symptoms associated with depression; Cluster #4 pertained to research methods; and Clusters #5 and #6 were related to the mechanisms of EA treatment for depression.

Figure 8. Cluster analysis of Chinese and English literature on electroacupuncture treatment for depression. (A) Cluster analysis diagram of Chinese literature; (B) Cluster analysis diagram of English literature.

3.6.3. Timeline View and Keyword Burstness Analysis

The clustered keyword maps were converted into timeline views (Figure 9 and Figure 10) to reflect the relationships among clusters and the temporal span of literature within each cluster. Keyword burstness reflects keywords that appear with high frequency within a certain period, indicating the importance of those terms in the research field and the changes in attention they receive over time. These two maps enable the exploration of research hotspots and the forecasting of research frontiers in the relevant field.

As inferred from Figures 9-11, in the included Chinese literature, antidepressants such as “fluoxetine” and “amitriptyline” garnered considerable attention between 2003 and 2010. Different types of EA, such as “music electroacupuncture” and “auricular electroacupuncture,” received heightened attention between 2010 and 2015. After 2015, keywords related to the mechanisms of EA treatment for depression, such as “neurotransmitter,” “gut microbiota,” and “metabolic pathway,” became research hotspots. In the included English literature, research hotspots from 2003 to 2010 primarily focused on animal models and antidepressants. Between 2010 and 2015, the microscopic mechanisms of EA treatment for depression began to attract researchers’ attention, with the emergence of keywords such as “hippocampus” and “BDNF”. Concurrently, clinical trial-related areas also gained attention, with the appearance of keywords such as “double-blind” and “randomized controlled”. After 2015, keywords such as “TRPV1” and “cannabinoid receptor 1” emerged, suggesting that research hotspots have primarily concentrated on the in-depth exploration of the microscopic mechanisms underlying EA treatment for depression.

Figure 9. Timeline chart of key terms in Chinese literature.

Figure 10. Timeline chart of key terms in English literature.

Figure 11. Highlighted keyword distribution in Chinese and English literature. (A) Keyword prominence map of Chinese literature; (B) Keyword prominence map of English literature.

Statistical analysis of annual publication volumes revealed that Chinese publications in this field have stabilized at a high level, while English publications have exhibited an overall upward trend. All core authors identified in both Chinese and English datasets are researchers based in China, which reflects that Chinese scholars dominate global publication output of electroacupuncture for depression research, with sustained stable annual output of Chinese-language literature.

Visualization analysis of author collaborations revealed that the authors in the included literature have formed a core author-driven team development model. The most influential domestic teams in this field are those led by Tu Ya, Li Zhigang, and Ren Lu. Meanwhile, the first publications of highly productive scholars in Fang Jiliang’s team and those involved in the English literature appeared relatively late, suggesting, to some extent, that the research directions of these teams are more closely aligned with recent research hotspots. Fang Jiliang’s team has focused on EA treatment for depression, specifically investigating brain mechanisms using fMRI following auricular EA and auricular concha EA in different populations with depression ([12]; [27]). Their recent research indicates that auricular concha EA has the effect of immediately modulating the functional activity of brain regions within the cognitive control network, default mode network, reward network, and visual processing network in patients with first-episode depression (FED), which may represent a potential brain effect mechanism underlying its treatment of FED ([23]).

The institutional collaboration analysis maps indicated that among the Chinese literature, Beijing University of Chinese Medicine had relatively close connections with other research institutions. Although institutions with relatively high publication volumes and influence have emerged in the map, inter-institutional connections remain limited. This lack of communication may lead to issues such as the inability to complement respective strengths ([8]). In the included English literature, domestic Chinese institutions accounted for a high proportion of the top 20 publishing institutions. Furthermore, the country collaboration map indicated that, apart from China, relatively few other countries are engaged in research in this field.

Existing clinical literature indexed in CNKI and WOS frequently reports positive therapeutic outcomes of electroacupuncture for depression, as reflected by the high frequency of clinical trial keywords in co-occurrence analysis ([2]; [11]), its global acceptance still needs improvement. Potential adverse effects of EA (such as hematoma and pain) ([28]), the relatively short duration of clinical trials on EA for depression, and individual variability in needling techniques ([30]) may contribute to this situation. This suggests that domestic scholars could strengthen collaborative relationships to promote the development of this field in China. Simultaneously, enhancing international exchange and cooperation could promote the internationalization of research in this field. Furthermore, scholars could conduct more in-depth reflection and research on clinical trials, employing more scientific and standardized treatment and evaluation criteria to improve international acceptance.

Analysis of keyword-related maps indicated that keywords such as “rat”, “chronic stress” and “fluoxetine” appeared with high frequency, suggesting that experimental designs related to depression in current research are relatively mature. Chronic unpredictable mild stress is the primary method for constructing animal models, and fluoxetine is frequently selected as the positive control drug. High-frequency keywords such as “clinical observation”, “quality of life” and “randomized controlled trial” indicate that clinically related research is also a key focus of this field.

Meanwhile, keywords related to different types of EA (e.g., “music electroacupuncture” and “auricular concha electroacupuncture”) and different acupoints stimulated by EA (e.g., “Baihui” (GV20) and “Yintang” (GV29)) appeared with high frequency. Corresponding cluster labels were present in the clustering maps, and these keywords have ceased to exhibit burstness (Figure 11), indicating that scholars have conducted considerable exploration in these areas in past research. Combining keywords from the clustering and timeline maps, such as “acetylcholine” and “metabolic pathway” cluster labels such as “TRPV1” and “cannabinoid receptor 1” and keywords still exhibiting burstness in the keyword burstness map, such as “mechanisms” it is evident that research on the mechanisms of EA treatment for depression has become a hotspot and focus of this field.

Regarding the efficacy of different EA types in treating depression, Ji Qian et al. found that both music EA and pulsed EA exerted antidepressant effects superior to fluoxetine, with music EA showing a trend toward superiority over pulsed EA in improving the behavior of rats with chronic stress-induced depression ([15]; [16]). Liu Yueting et al. ([17]) discovered that both auricular margin EA and auricular concha EA improved depressive behavior in rats, but the efficacy of auricular margin EA was inferior to that of auricular concha EA. Regarding the efficacy of EA stimulation at different acupoints for treating depression, Roberto Gonzalez et al. found ([9]), through intervention in different populations with depression, that EA at Baihui (GV20) and Yintang (GV29), combined with manual acupuncture at Sanyinjiao (SP6), effectively controlled and reduced depressive symptoms. Chen Tao et al. ([4]) suggested that EA at the Siguan acupoints significantly improved behavioral indicators in rats with post-stroke depression. Du Li et al. proposed that EA at Shenmen (HT7) - Sanyinjiao (SP6) was superior to EA at Sanyinjiao (SP6) alone in improving insomnia severity and anxiety/depression status, with Shenmen (HT7) playing a major role in alleviating anxiety and depression ([7]).

Among Chinese publications in 2023, Lu Yujia et al. suggested that the mechanism by which EA ameliorates depressive symptoms may be related to the regulation of BDNF/TrkB/CREB signaling pathway-related proteins in hippocampal tissue and the elevation of monoamine neurotransmitters 5-HT, NE, and DA levels ([18]). Song Guixia et al. proposed that EA combined with fluoxetine improves depressive behavior, potentially through a mechanism involving the improvement of the diversity and relative abundance of beneficial gut microbiota and the upregulation of brain-gut axis-related molecule expression, such as 5-HT and 5-HTP ([22]). Among English publications in 2023, [14] suggested that EA ameliorates post-stroke depression (PSD)-related behavioral and cognitive dysfunction and promotes mitochondrial biogenesis by activating CB1R, representing a novel mechanism of EA treatment for PSD. [20] indicated that P2X7R might be a target for EA intervention in the liver-brain axis for treating depression.