Traditional Chinese medicine therapies for insomnia: An umbrella review and evidence map

Abstract

Background

An increasing number of systematic reviews and meta-analyses (SR/MAs) suggests traditional Chinese medicine therapies are effective for insomnia. We aimed to synthesize and evaluate the methodological quality of these studies through an umbrella review with an evidence map for improving evidence quality.

Methods

We searched 10 databases from inception to March 20, 2025, that investigated the effects of TCM therapies on sleep-related subjective or objective outcomes for insomnia. We assessed the methodological quality of included SR/MAs using A Measurement Tool to Assess systematic Reviews (AMSTAR 2) tool, evaluated the certainty of evidence with the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) tool, and visually presented the results as an evidence map.

Results

Thirty-six SR/MAs included in this umbrella review described 3 TCM non-pharmacotherapies (i.e., acupuncture, Tuina massage, and Chinese exercises), and Chinese herbal medicine (CHM). The AMSTAR 2 results showed that 20 SR/MAs had high or moderate methodological quality, while the remaining studies were of low or critically low methodological quality. The evidence map showed high-quality SR/MAs supported the effect of acupuncture, Tuina massage, Chinese exercises, and CHM on overall sleep quality, while acupuncture, Tuina massage, and Chinese exercises could also improve anxiety and depression. Moreover, acupuncture and Tuina massage might improve objective outcomes such as polysomnography parameters and 5-hydroxytryptamine level.

Conclusions

Acupuncture, Tuina massage, and Chinese exercises are effective in improving overall sleep quality and emotional outcomes and have potential effects on objective sleep parameters. There is sufficient evidence that several CHM formulas could improve sleep quality. However, the methodological quality of SR/MAs needs further improvement.

Protocol registration

PROSPERO, CRD42022347769.

1. Introduction

Insomnia is a highly prevalent disorder characterized by difficulty in initiating and/or maintaining sleep, alongside daytime dysfunction, despite adequate opportunities for sleep.¹ The global prevalence of insomnia in adults in the past 3 years has reached 27.29 %.² Its symptoms can be worse in some vulnerable population groups, such as health care workers and older adults.3, 4, 5 Insomnia may lead to poor quality of life and poor physical performance, and may increase morbidity and mortality from cardiovascular diseases.^6,7 Moreover, individuals with untreated insomnia each year have a 63,607 USD higher all-cause economic cost than good sleepers.⁸

Cognitive behavioral therapy for insomnia (CBT) is recommended as the first-line treatment. However, the high cost and insufficient number of experienced practitioners have limited its application globally.¹ Pharmacological therapies should be considered if CBT fails or is unavailable, but they are primarily for short-term insomnia and may cause daytime drowsiness and increased risks of falls and dependence.9, 10, 11 Due to the limitations of these therapies, complementary and alternative therapies such as traditional Chinese medicine (TCM) therapies are becoming another choice for the treatment of insomnia.

Recommendations of TCM therapies for insomnia in current clinical practice guidelines are weak due to the lack of high-quality systematic reviews and meta-analyses (SR/MAs) and robust evidence.¹² In the past few years, several SR/MAs of randomized controlled trials (RCTs) have tried to confirm that TCM therapies are beneficial for insomnia.13, 14, 15 However, some RCTs that they included had a high risk of bias, and some SR/MAs had poor methodological quality.16, 17, 18 Therefore, it is necessary to evaluate the methodological quality of the current SR/MAs, assess the certainty of evidence, and categorize them with an evidence map according to their quality and conclusions. We aimed to conduct an umbrella review on the effect of TCM therapies for insomnia and systematically evaluate the certainty of evidence, which could provide researchers, health care providers, and policymakers with robust evidence and also lay foundations for the future development of related guidelines.

2. Methods

2.1. Registration of the review

We reported this study following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) and Preferred Reporting Items for Overviews of reviews (PRIO) checklists.^19,20 The protocol of this study was registered on the International Prospective Register of Systematic Reviews (PROSPERO): CRD42022347769.

2.2. Search strategy

We conducted a comprehensive literature search for SR/MAs in the following databases from their inception to March 20, 2025: PubMed, Web of Science, EMBASE, the Cochrane library, PsycINFO, PubPsych, Chinese Journal Full-text Database, Wanfang, SinoMed, and Chongqing VIP. A forward and backward search of the reference lists in the eligible studies, Google Scholar, and other internet search engines was performed. The full search strategies are available in Supplement 1.

2.3. Eligibility criteria

The eligibility criteria were developed according to the PICO-S (population, intervention, comparison, outcomes, study design) framework.

Population: We included SR/MAs that targeted patients with insomnia as per standard criteria [e.g., DSM-5 (Diagnostic and Statistical Manual of Mental Disorders, fifth edition), ICD-11 (International Classification of Diseases, 11th edition), and CCMD-3 (Chinese Classification of Mental Disorders, third edition)].21, 22, 23 There were no restrictions on age, gender, ethnicity, or comorbidities.

Intervention: TCM therapies were applied in the intervention group. Different TCM therapies could be synthesized in one meta-analysis. If a meta-analysis included primary studies with two or more TCM therapies in the intervention group, it was excluded.

Comparison: Patients in the control group received standard treatment such as western medicine, CBT, and/or placebo treatment.

Outcomes: We focused on outcomes directly or indirectly related to sleep and categorized them as follows: (1) questionnaires measuring the patients’ sleep conditions, defined as patient-reported sleep-related outcomes [e.g., Pittsburg Sleep quality Index (PSQI) or Insomnia Severity Index (ISI)]; (2) questionnaires measuring the patients’ conditions that might affect sleep, such as depression and anxiety, defined as patient-reported indirect sleep-related outcomes [e.g., Hamilton Depression/Anxiety Scale, (HAMD/HAMA); Self-rating Anxiety/Depression Scale, (SAS/SDS)]²⁴; and (3) objective outcomes [e.g., polysomnography (PSG) parameters], defined as objective sleep-related outcomes.

Study design: SR/MAs of RCTs were eligible. If the conclusions of several SR/MAs with the same PICO framework were the same, we chose the SR/MA with the largest sample size; otherwise, all these SR/MAs were included.25, 26, 27 Only English or Chinese studies were considered, with no other restrictions.

2.4. Study selection and data extraction

The retrieved studies were imported into EndNote 20.3, and duplicates were found by the function “Find Duplicate.” Two reviewers (WJX and BB) independently screened the titles and abstracts for studies that might be included. After that, full-text evaluation was performed for further selection. The final results were cross-checked, and discrepancies were solved by discussion or judged by FL.

Data extraction was performed by two reviewers independently (GCY and TXC). Discrepancies were solved by discussion or by consulting FL to reach consensus. We extracted the following data from the included SR/MAs: name of the first author and the publication year, number of trials included and the sample size, population, type of TCM therapy, control methods, outcomes, author’s conclusion, and effect size of each comparison (with the corresponding confidence intervals, P values, and the I²).

2.5. Quality assessment

We used A MeaSurement Tool to Assess systematic Reviews (AMSTAR 2) to assess the methodological quality of the eligible SR/MAs. The AMSTAR 2 tool consists of 16 items, which targets the literature search, eligibility criteria, study selection and data extraction, quality appraisal of included studies, statistical analysis and conflict of interest of SR/MAs. Among them, seven items are regarded critical (i.e., item 2, 4, 7, 9, 11, 13, and 15). For each item, the answer could be “yes”, “partially yes”, “no”, or “not applicable". Full descriptions of all 16 items were listed in Supplement 2. The methodological quality of an SR/MA can be classified as high (no or only one non-critical weakness), moderate (more than one non-critical weakness but no critical flaws), low (one critical flaw with or without non-critical weakness), or critically low (more than one critical flaw with or without non-critical weaknesses).²⁸

We extracted meta-analytic comparison results from included SR/MAs, and evaluated the certainty of evidence using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) tool. Meta-analyses of RCTs are regarded as high-certainty evidence, which can be rated down if there are risk of bias, inconsistency, indirectness, imprecision, and publication bias in the included studies. The overall certainty of evidence is rated as high, moderate, low, or very low.²⁹ The quality assessment were performed by two independent reviewers, with discrepancies solved by consulting LF.

2.6. Evidence map

We displayed our findings through an evidence map. The horizontal axis represented the directions of TCM therapy’s effect, which were categorized with “uncertain”, “possibly effective, and “effective” according to conclusions of the included SR/MAs. The vertical axis represented the methodological quality of the included SR/MAs, which were categorized with “high”, “moderate”, “low”, and “critically low" based on the AMSTAR 2 assessment results. Each bubble represented one SR/MA. Larger bubble size indicated larger sample size.

3. Results

3.1. Search results

We identified 3184 records from the online search and reduced them to 2469 after removing duplicates. The full text of 73 studies were evaluated, and ultimately, 36 SR/MAs that met the inclusion criteria were included (Fig. 1).^15,30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65 Reasons for the exclusion of full-text studies are available in Supplement 3. The eligible SR/MAs were published between 2015 and 2025. The number of primary studies included in these SR/MAs ranged from 8 to 109, with the sample size ranging from 650 to 11,488. Patients with primary insomnia or those with comorbidities such as stroke, chronic pain, anxiety, depression, and cancer were included (Table 1).

Fig. 1.

Study flow diagram.

Table 1.

Characteristic of the included studies.

First author (Year)	No. of trials (sample size)	Type of insomnia	Interventions	Control	Outcomes	Author’s conclusion	Methodological quality (AMSTAR 2)
Lan (2015)55	15 (1429)	primary	AA	sham AT	PSQI	±	Low
Ni (2015)15	79 (7886)	NS	CHM	WM	PSQI	±	Low
Jin (2016)46	14 (1207)	NS	CHM (Guipi D)	WM	PSQI	+	Moderate
Li (2016)61	11 (1075)	elderly	AT	WM	PSQI	±	Critically low
Jiang (2017)30	14 (1182)	NS	Baduanjin	wait-list, sleep education, WM	1) PSQI; 2) SAS	1) +; 2) ±	Critically low
Kui (2019)42	61 (5128)	primary	AT	WM	1) PSQI; 2) SDS; 3) SAS	1)-3) ±	High
Liu (2019)53	9 (944)	chronic pain	MA	sham AT, WM	PSQI	+	Low
Yang (2019)49	14 (1256)	NS	CHM (TWBXD)	WM	PSQI	-	Low
Birling (2020)57	19 (1780)	NS	CHM (ZRAS)	placebo, BzRAs	PSQI	-	Low
Li (2020)41	14 (1549)	NS	CHM (YXAS)	WM, placebo	1) PSQI; 2) PSG	1)-2) +	High
Li (2020)40	34 (2665)	NS	ACE	estazolam	PSQI	+	High
Wang (2020)45	22 (2029)	NS	CHM (CLMD)	WM	PSQI	±	Moderate
Fang (2020)44	9 (617)	NS	AT	WM	1) PSQI; 2) PSG	1) +; 2) ±	High
Hu (2021)43	9 (681)	with anxiety	CHM (XYS)	WM	1) PSQI; 2) HAMA; 3) SAS	1)-3) +	High
Lin (2021)54	14 (910)	NS	CHM (Banxia)	WM	PSQI	+	Low
Liu (2021)60	21 (1606)	NS	SA	WM	PSQI	+	Critically low
Pei (2021)39	8 (618)	with hemodialysis	AA	estazolam	PSQI	+	High
Peng (2021)59	34 (2876)	perimenopausal	AT -moxibustion	WM	PSQI	+	Critically low
Tan (2021)52	18 (1573)	NS	CHM (XYS)	WM	PSQI	+	Low
Zhang (2021)35	44 (3663)	primary	AT	WM	1) PSQI; 2) ISI; 3) SAS; 4) HAMD	1)-4) +	High
Zhang (2021)58	35 (3200)	NS	CHM (Suanzaoren)	WM	PSQI	+	Critically low
Zhao (2021)33	11 (775)	primary	AT	sham AT, wait-list	1) PSQI; 2) ISI; 3) PSG	1)-3) +	High
Chen (2022)47	15 (1164)	NS	CHM (GGLMD)	WM	PSQI	+	Moderate
Han (2022)56	21 (2022)	NS	Tai Chi	conventional exercise, no treatment, etc.	PSQI	+	Low
Wu (2022)51	8 (650)	NS	Baduanjin	CHM, walking, WM, etc.	1) PSQI; 2) SAS	1)-2) +	Low
Yu (2022)36	13 (1109)	cancer-related	AT	sham AT, wait-list, WM	1) PSQI; 2) ISI	1) +; 2) ±	High
Zhao (2022)34	21 (1571)	with depression	AT	WM	1) PSQI; 2) HAMD; 3) SDS	1)-3) +	High
Zhou (2022)32	19 (1850)	NS	CHM (Wuling)	WM	PSQI	+	High
Zhou (2022)31	26 (1874)	post-stroke	AT	WM	PSQI	+	High
Yang (2023)37	16 (1547)	NS	Tai chi	conventional exercise, no treatment	1) PSQI; 2) HAMD; 3) HAMA; 4) SAS	1)-4) +	High
Wang (2024)38	23 (1780)	NS	tuina massage	WM, CHM	1) PSQI; 2) SAS; 3) SDS	1)-3) +	High
Zhang (2024)48	16 (1309)	with hypertension	AT	WM	PSQI	+	Low
Su (2024)63	54 (4553)	post-stroke	AT or CHM	sham/placebo, WM	PSQI	+	High
Li (2025)65	50 (4226)	NS	Five Chinese exercise	usual care	PSQI	+	Low
Ma (2025)62	109 (11,488)	Primary	Chinese patent medicine	WM, placebo	1) PSQI; 2) PSG	1)-2) +	High
Zhao (2025)64	14 (1207)	with depression	CHM (Shumian)	antidepressants	1) PSQI; 2) HAMD	1) +; 2) ±	High

AA, auricular acupuncture; ACE, acupoint catgut embedding; AT, acupuncture; BzRAs, Benzodiazepine Receptor Agonists; CHM, Chinese herbal medicine; CLMD, Chaihu Longgu Muli decoction; Guipi D, Guipi decoction; GGLMD, Guizhi Gancao Longgu Muli decoction; HAMA, Hamilton anxiety scale; HAMD, Hamilton depression scale; ISI, insomnia severity index; MA, manual acupuncture; NS, not specified; PSG, polysomnography; SA, Scalp acupuncture; SAS, self-rating anxiety scale; SDS, self-rating depression scale; SRSS, self-rate sleep scale; TWBXD, Tian Wang Bu Xin Dan; XYS, Xiao Yao San; WM,western medicine; ZRAS, Zao Ren An Shen;.

+, positive; ±, neutral; -, negative or uncertain.

3.2. Methodological quality results

The AMSTAR 2 appraisal identified 17 high–quality SR/MAs, 3 moderate–quality SR/MAs, 11 low–quality SR/MAs, and 5 critically low–quality SR/MAs (Supplement 4). In terms of critical items, 13 studies did not have a protocol or registration (item 2); 9 studies did not search gray literature or the reference list (item 4); 33 studies did not list the excluded full-text studies, and 5 of them did not even give reasons for this in the flow diagram or text (item 7); 6 studies did not present sufficient consideration when interpreting the results (item 13); and 2 studies did not provide a full discussion regarding whether the potential risk of bias affected the results of the review (item 15). All studies met the requirement of item 9 and 11. In terms of non-critical domains, 13 studies did not give sufficient characteristic data of the included studies (item 8); 4 studies did not fully consider the impact of risk of bias in individual studies on the results (item 12); 12 studies did not fully consider the impact of heterogeneity on the results (item 14); and 4 studies did not report any potential conflicts of interest or the funding information (item 16). All studies met the requirement of item 1, 3, 5, and 6, but they did not report source of funding of their included studies (item 10).

As shown in Fig. 2, 9 high-quality SR/MAs drew positive conclusions that acupuncture was effective (n = 5) or possibly effective (n = 4) for insomnia. Six high-quality and 4 moderate-quality SR/MAs found that CHM was effective (n = 6) or possibly effective (n = 4) for insomnia. Only one high-quality study on Tuina massage was included, which showed that Tuina massage was effective for insomnia. Five included SR/MAs showed that Chinese exercises were effective for insomnia, but only one study had high methodological quality.

Fig. 2.

Evidence map of Traditional Chinese medicine therapies for insomnia.

The horizontal axis indicates the direction of the original authors’ conclusions regarding the effects of TCM therapies, while the vertical axis denotes review quality. Bubble size corresponds to sample size. The bubble representing Su et al.'s study is divided into two colors, reflecting separate meta-analyses for Chinese herbal medicine and acupuncture.

3.3. Certainty of evidence

A total of 76 different comparisons were identified (Table 2). GRADE showed that 19 of them were supported by moderate-certainty evidence, 42 of them were supported by low-certainty evidence, and the remaining 15 comparisons were supported by very-low-certainty evidence. The authors of the SR/MAs argued that some of the primary studies they included had low quality due to implicit allocation concealment and the blinding of outcome assessors. Moreover, the heterogeneity between trials in some comparisons was large, which also affected the certainty of evidence.

Table 2.

Meta-analysis results and certainty of evidence.

Study	Comparison	Outcome	No. of trials (treatment / control)	Certainty of evidence	Effect size (95 % CI)
Lan (2015)55	AA vs sham / placebo AA	PSQI	5 (337/324)	Moderate	MD: −3.41 (−3.93, −2.89)
Ni (2015)15	CHM vs WM	PSQI	35 (1766/1595)	Low	MD: −1.94 (−2.45 to −1.43)
Jin (2016)46	CHM (Guipi) vs WM	PSQI	3 (171/157)	Very low	MD: −4.52 (−7.67 to −1.36)
Li (2016)61	AT vs WM	PSQI	6 (238/236)	Low	MD: −1.36 (−1.93 to −0.79)
Jiang (2017)30	Baduanjin vs wait-list, sleep education, WM, etc.	PSQI	13 (556/561)	Very low	SMD: −3.78 (−5.09 to −2.47)
		SAS	2 (70/75)	Very low	SMD: −1.28 (−6.07 to 3.52)
Kui (2019)42	AT vs WM	PSQI	40 (1715/1648)	Low	MD: −1.28 (−3.00 to −1.59)
		SDS	4 (132/127)	Very low	MD: −5.17 (−11.33 to 0.99)
		SAS	2 (69/66)	Low	MD: −3.30 (−5.69 to −0.91)
Liu (2019)53	MA vs WM	PSQI	6 (353/353)	Low	MD: −2.44 (−3.84 to −1.05)
Yang (2019)49	CHM (TWBXD) vs WM	PSQI	5 (292/182)	Low	MD: −1.82 (−3.00 to −0.64)
Birling (2020)57	CHM (ZRAS) vs WM	PSQI	8 (321/321)	Low	MD: 0.17 (−0.29 to 0.64)
Li (2020)41	CHM (YXAS) vs WM	PSQI	2 (126/126)	Low	MD: −1.16 (−3.80 to 1.49)
	CHM (YXAS) vs placebo	PSG (TST)	2 (126/128)	Moderate	MD: 5.16 (3.09 to 7.23)
		PSG (SE)	2 (126/128)	Moderate	MD: 1.21 (0.80 to 1.61)
		PSG (SOL)	2 (126/128)	Moderate	MD: −2.87 (−4.12 to −1.62)
		PSG (WASO)	2 (126/128)	Moderate	MD: −2.28 (−4.07 to −4.09)
Li (2020)40	ACE vs estazolam	PSQI	17 (488/497)	Low	SMD: −1.10 (−1.49 to −0.71)
Wang (2020)45	CHM (CLMD) vs WM	PSQI	10 (535/534)	Low	MD: −2.72 (−5.00 to −0.44)
Fang (2020)44	AT vs WM	PSG (TST)	2 (85/85)	Low	MD: −6.35 (−16.07 to 3.37)
		PSG (SE)	1 (45/45)	Low	MD: −0.03 (−0.87 to 0.81)
		PSG (SOL)	2 (85/85)	Low	MD: −0.68 (−2.33 to 0.98)
		N1 sleep	1 (45/45)	Low	MD: −7.65 (−8.91 to −6.39)
		N2 sleep	1 (45/45)	Low	MD: −10.7 (−12.54 to −8.86)
		N3 sleep	1 (45/45)	Low	MD: 9.43 (8.31 to 10.55)
		REM sleep	1 (45/45)	Low	MD: 8.92 (6.48 to 11.36)
Hu (2021)43	CHM (XYS) vs WM	PSQI	4 (126/126)	Low	MD: −4.45 (−6.65 to 2.24)
		HAMA	2 (60/60)	Very low	MD: −5.19 (−7.78 to −2.60)
		SAS	3 (133/132)	Low	MD: −9.38 (−10.20 to −8.57)
Lin (2021)54	CHM (Banxia) vs WM	PSQI	6 (172/164)	Moderate	MD: −1.05 (−1.63 to −0.47)
Liu (2021)60	SA vs WM	PSQI	9 (354/356)	Moderate	MD: −1.49 (−2.05 to −0.92)
Pei (2021)39	AA vs estazolam	PSQI	2 (50/50)	Very low	MD: −1.15 (−4.97 to 2.68)
Peng (2021)59	AT vs WM	PSQI	27 (1115/1110)	Low	MD: −2.09 (−2.91 to −1.27)
Tan (2021)52	CHM (XYS) vs WM	PSQI	3 (114/110)	Very low	MD: −2.25 (−3.37 to −1.12)
Zhang (2021)35	AT vs WM	PSQI	31 (1284/1224)	Very low	MD: −2.09 (−2.62 to −1.57)
		ISI	3 (104/103)	Very low	MD: −1.78 (−3.24 to −0.32)
		SAS	4 (143/142)	Very low	MD: −4.62 (−7.44 to −1.81)
		HAMD	3 (99/97)	Very low	MD: −3.74 (−7.49 to 0.01)
Zhang (2021)58	CHM (Suanzaoren) vs WM	PSQI	16 (765/759)	Low	MD: −3.99 (−4.59 to −3.39)
		SRSS	7 (315/312)	Low	MD: −5.17 (−5.55 to −4.79)
Zhao (2021)33	AT vs sham AT / wait-list	PSQI	9 (350/289)	Moderate	MD: −4.23 (−5.52 to −2.93)
		ISI	4 (198/134)	Low	MD: −3.94 (−5.94 to −1.95)
		PSG (TST)	11 (418/357)	Moderate	MD: 55.29 (29.16 to 81.42)
		PSG (SE)	10 (393/334)	Moderate	MD: 8.96 (3.97 to 13.95)
		PSG (NAT)	7 (231/236)	Moderate	MD: −6.29 (−10.75 to −1.82)
		PSG (WASO)	7 (225/230)	Moderate	MD: −49.54 (−83.0 to −16.1)
Chen (2022)47	CHM (GGLMD) vs WM	PSQI	6 (249/249)	Low	MD: −2.70 (−3.85 to −1.54)
Han (2022)56	Tai chi vs conventional exercises or no treatment	PSQI	21 (907/928)	Low	MD: −1.16 (−1.62 to −0.71)
Wu (2022)51	Baduanjin vs usual care, CHM, walking, or WM	PSQI	5 (182/179)	Low	MD: −2.20 (−2.89 to −1.52)
		SAS	2 (82/79)	Very low	MD: −3.87 (−5.67 to −2.07)
Yu (2022)36	AT vs sham AT, wait-list, WM	PSQI	10 (416/415)	Low	MD: −1.83 (−2.71 to −0.94)
		ISI	4 (120/117)	Low	MD: 0.16 (−1.92 to 2.23)
Zhao (2022)34	AT vs WM	PSQI	10 (333/335)	Low	MD: −1.17 (−2.26 to −0.08)
		HAMD	7 (247/249)	Very low	MD: −0.47 (−0.91 to −0.02)
		SDS	3 (86/86)	Very low	MD: 2.10 (−4.20 to 8.39)
Zhou (2022)32	CHM (Wuling) vs WM	PSQI	18 (878/868)	Low	MD: −1.92 (−2.34 to −1.50)
		SDRS	4 (157/153)	Low	MD: −4.21 (−4.95 to −3.46)
Zhou (2022)31	AT vs WM	PSQI	22 (807/799)	Very low	MD: −3.41 (−4.41 to −2.40)
Yang (2023)37	Tai chi vs conventional exercises, no treatment	PSQI	16 (781/766)	Moderate	MD: −2.05 (−2.42 to −1.68)
		HAMA	2 (60/60)	Moderate	MD: −2.18 (−2.98 to −1.37)
		HAMD	3 (95/95)	Moderate	MD: −5.08 (−5.46 to −4.69)
		SAS	2 (194/196)	Moderate	MD: −7.01 (−7.22 to −6.29)
Wang (2024)38	Tuina massage vs WM, CHM	PSQI	14 (489/483)	Low	MD: −2.34 (−2.74 to −1.94)
		SAS	9 (380/372)	Low	MD: −6.77 (−8.34 to −5.20)
		SDS	7 (323/39)	Low	MD: −6.60 (−8.82 to −4.37)
Zhang (2024)48	AT vs WM	PSQI	9 (377/342)	Low	MD: −3.20 (−3.77 to −2.62)
Su (2024)63	CHM vs BZDs	PSQI	15 (1128)	Moderate	MD: −2.42 (−2.97 t −1.87o)
	CHM vs non-BZDs	PSQI	3 (153)	Low	MD: −0.57 (−3.08 to 1.93)
	CHM + BZDs vs BZDs	PSQI	4 (313)	Low	MD: −2.14 (−3.03 to −1.26)
	AT vs BZDs	PSQI	8 (693)	Moderate	MD: −2.01 (−3.05 to −0.98)
	AT + BZDs vs BZDs	PSQI	4 (295)	Low	MD: −3.38 (−4.24 to −2.53)
	AT + non-BZDs vs non-BZDs	PSQI	3 (287)	Moderate	MD: −2.85 (−3.63 to −2.07)
Li (2025)65	Five Chinese exercise vs usual care	PSQI	50 (4226)	Low	network meta-analysis
Ma (2025)62	Chinese patent medicine vs WM	PSQI, TST, SOL	109 (11,488)	Low	network meta-analysis
Zhao (2025)64	Chinese patent medicine vs	PSQI	7 (667)	Low	MD: −1.53 (−1.95 to −1.11)
	CHM (Shumian) vs antidepressant	HAMD	13 (1127)	Low	MD: −0.92 (−1.53 to −0.31)

AA, auricular acupuncture; ACE, acupoint catgut embedding; AT, acupuncture; BZDs, Benzodiazepines; CHM, Chinese herbal medicine; CLMD, Chaihu Longgu Muli decoction; GGLMD, Guizhi Gancao Longgu Muli decoction; HAMA, Hamilton anxiety scale; HAMD, Hamilton depression scale; ISI, insomnia severity index; MA, manual acupuncture; NAT, no. of awakening times; PSG, polysomnography; SA, Scalp acupuncture; SAS, self-rating anxiety scale; SDS, self-rating depression scale; SRSS, self-rate sleep scale; TWBXD, Tian Wang Bu Xin Dan; XYS, Xiao Yao San; WM,western medicine; ZRAS, Zao Ren An Shen.

4. Discussion

4.1. Summary of main results

This umbrella review appraised the methodological quality of 36 SR/MAs on TCM therapies for insomnia, and visually displayed the effects of identified treatments through an evidence map. The AMSATR 2 assessment results showed 20 studies were rated as having high or moderate quality, while the remain 16 studies were of low or critically quality. According to the evidence map, high-quality studies showed acupuncture, Tuina massage, Chinese exercises, and several CHM formulas could improve sleep quality.

Eight high-quality SR/MAs proposed that acupuncture, especially manual acupuncture, was beneficial for sleep outcomes compared with western medicine or sham acupuncture, as it improved sleep quality, reduced insomnia severity, relieved anxiety, and improved sleep structure and PSG parameters (e.g., sleep onset latency and total sleep time). Besides, the between-group difference in sleep quality was detected since the second week of treatment with manual acupuncture.⁴² One study also presented data on the effects of acupoint catgut embedding (ACE) on sleep quality and concluded that ACE was more effective than estazolam in improving long-term sleep quality.⁴⁰ We identified 15 SR/MAs that investigated the effect of Chinese herbal medicine (CHM) on insomnia. Most studies used different CHM modalities, which were regarded as different treatments. Among them, three studies gave a whole picture of the general combined effects of various CHM for insomnia that CHM could improve overall sleep quality.^15,62,63 In terms of a single CHM formula for insomnia, high- or moderate-quality studies showed that Xiao Yao San, Wuling Capsule, Chaihu Longgu Muli decoction, Guizhi Gancao Longgu Muli decoction, and Guipi decoction could improve subjective sleep quality. The effect and safety of Tuina massage on improving sleep quality, relieving anxiety and depression states, and increasing 5-hydroxytryptamine (5-HT) levels have been demonstrated by a high-quality SR/MA.³⁸ 5-HT is one of the monoamine neurotransmitters that are related to sleep–wake cycle, and its decreased levels may cause insomnia. Tuina massage can stimulate the skin and deep tissue receptors, generate neural signals that are sent to the brain, modulate the hypothalamic–pituitary–adrenal axis, and finally increase 5-HT levels. Our umbrella view found Chinese exercises are generally effective to overall sleep quality.⁶⁵ One high-quality SR/MA suggested that Tai Chi, the most well-known Chinese exercise, could improve sleep quality as well as relieve negative emotions such as anxiety and depression. Tai Chi may improve the function and connectivity of sleep-related brain structures, including the hippocampus, prefrontal cortex, orbitofrontal cortex, and caudate nucleus, thereby treating insomnia.^66,67 Of note, the control group in this SR/MA included western medicine, conventional exercises, and acupuncture, which might be a source of heterogeneity. Future SR/MAs need to make subgroup analysis to make the results precise, and we should interpret and apply the current evidence with caution.

4.2. Agreement and disagreements with other studies or reviews

This study corroborates the findings of two prior umbrella reviews demonstrating acupuncture's efficacy in improving sleep quality, while strengthening the robustness of their conclusions. For instance, Ell et al. based their conclusions without evaluating Chinese-language articles, potentially introducing selection bias given that nearly half of global acupuncture studies originate from China.^68,69 Similarly, He et al. included 34 SR/MAs, yet did not account for study overlap.¹⁸ This could cause overestimation of acupuncture effects as 30 SR/MAs in He’s review were from China. To address these limitations, we conducted a comprehensive literature search of 3 Chinese databases and the reference list of eligible studies. Only the SR/MA with the largest dataset (prioritized by PICO framework and conclusion direction) was included to minimize redundancy. Our findings on Chinese exercises (a specific mind-body aerobic regimen characterized by low-to-moderate intensity movements) align with a previously published review on exercise interventions for insomnia.⁷⁰ We observed that these exercises significantly improve sleep quality.

4.3. Implication for clinical practice and research

This study aimed at providing robust evidence of TCM therapies for insomnia for clinical practitioners and researchers by evaluating and mapping SR/MAs on this topic. Four sorts of TCM therapies (i.e., acupuncture, CHM, Tuina massage, and Chinese exercises) were identified, which laid the foundation for forming strong TCM recommendations in future clinical practice guidelines. Twenty studies published in the most recent 5 years were rated as having high or moderate quality by the AMSTAR 2 tool, indicating that authors in this field have gradually noticed that a standardized and scientific process is the foundation for more reliable and convincing conclusions. However, there were also 16 studies rated as low or critically low quality. None of these reported the funding information of the included studies. Authors of SR/MAs are encouraged to report this information, especially if they include studies that were funded commercially, which have larger possibility for publication bias. Fourteen studies did not mention any registered protocol, indicating that some SR/MAs of TCM therapies for insomnia might have selectively chosen and reported studies or outcomes with positive results. In addition, some studies did not perform an additional literature search of the reference list or websites, and did not cite the excluded full-text articles. These flaws could make authors omit eligible studies, and intentionally select articles with positive results. Some high-quality SR/MAs have drawn on low-certainty evidence resulted from including low-quality randomized controlled trials, which has already affected the strength of recommendations in current clinical practice guidelines as well as hindered clinical practice of TCM therapies.¹² Future researchers should set more strict selection criteria for high-quality RCTs when updating the evidence, such as excluding studies without proper allocation concealment or blinding, rather than simply adding the latest studies. In summary, based on the evidence appraisal process of this study, we suggest researchers report their future work following both the PRISMA and AMSTAR 2 checklist to further improve the quality and make the conclusions more reliable.

4.4. Strength and limitations of this study

This review has several strengths, including a comprehensive literature search of different types of TCM therapies for insomnia from eight databases, a strict appraisal process using the AMSTAR 2 and the GRADE tool by two independent reviewers, the involvement of both subjective and objective outcomes, and the use of an evidence map to visually summarize and present the findings. However, we used the quality assessment results of primary studies as reported in the included SR/MAs instead of conducting a reassessment process. The quality of most of the RCTs in the SR/MAs was low due to poor allocation concealment and blinding, so it was difficult for us to perform sensitivity analysis for the results of the included SR/MAs by excluding low-quality studies. Moreover, we only searched English and Chinese databases, so any studies published in other languages may have been missed.

In conclusion, high-quality SR/MAs suggest that acupuncture, Tuina massage, and Chinese exercises are efficacious in improving sleep quality and anxiety-depression emotions, and some CHM formulas could also improve sleep quality. The conclusions were partially affected by methodological quality and certainty of evidence due to poor research processes in some studies. More high-quality SR/MAs of well-designed RCTs are needed to enhance the overall interpretability of our results for the further clinical application of TCM therapies in this field.

Declaration of competing interest

The authors declares that there is no conflict of interest in the publication of this article.

Funding

This study was funded by grant ZY(2021-2023)-0212 from the Three-year Action Plan for the Development of Traditional Chinese Medicine in Shanghai-Highland.

Ethical statement

No ethical approval was required as this study did not involve human participants or laboratory animals.

CRediT authorship contribution statement

Jinxiang Wang: Validation, Formal analysis, Writing – original draft, Writing – review & editing. Bing Bai: Validation, Formal analysis, Writing – original draft, Writing – review & editing. Ranran Zhu: Methodology, Visualization. Xintong Yu: Methodology, Visualization. Xiaoting Xu: Funding acquisition, Supervision. Xiaomin Tu: Methodology, Visualization. Lei Fang: Conceptualization, Writing – review & editing, Project administration, Supervision.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.