The Role of Complementary and Alternative Medicine on Cancer-Related Fatigue in Adults: An Overview of Systematic Reviews

Abstract

Background:

Cancer-related fatigue (CRF) has an enormous adverse impact on quality of life and subsequent therapy of cancer patients. Complementary and alternative medicine (CAM) is reported to improve CRF in many systematic reviews (SRs), but the effects are controversial because of variations in the quality and outcomes.

Methods:

Thirteen databases were searched from inception to September 2022. Only SRs of randomized controlled trials (RCTs) were included. We assessed the quality of included SRs with the AMSTAR-2 tool, the strength of evidence with the GRADE system, the risk of bias with the ROBIS tool, and the integrity of SRs with the PRISMA checklist.

Results:

We included 30 eligible SRs (27 meta-analyses). Based on the AMSTAR-2 tool, 29 SRs were rated as “critically low” quality, and only one was rated as “low” quality. With the ROBIS tool, 19 SRs demonstrated a low risk of bias. According to the PRISMA checklist, no SRs reported all the items, and 10 SRs sufficiently reported over 70%. Based on the GRADE system, 7 outcomes were assessed as high-quality evidence.

Conclusion:

This overview demonstrates promising evidence for the effectiveness of CAM interventions in the treatment of CRF in adults. The roles of qigong, music, auricular point therapy, and dietary supplements in CRF need further evaluation. Although findings are mixed, it is recommend to select appropriate CAM to manage cancer-related fatigue under the guidance of physicians. More studies with rigorous methodological designs and sufficient sample sizes are needed.

Introduction

Cancer seriously endangers human health. The latest statistics show that 19.3 million new cases and 10.0 million cancer deaths are expected worldwide in 2020. ¹ Given cancer itself and the advanced treatments, many symptoms occur in cancer patients and survivors. Among them, fatigue is a common and universal symptom.

Cancer related fatigue (CRF), caused by cancer or cancer treatment, is perceived as a distressing, persistent sense of tiredness or exhaustion characterized by physical, emotional, and/or cognitive fatigue. ² It has an enormous adverse impact on the usual functioning, quality of life (QoL), and subsequent therapy of cancer patients, which leads to poor prognosis. According to a meta-analysis involving 144 813 patients, the prevalence of CRF was 52%. ³ In addition, other reports have shown that 80% of cancer patients after chemotherapy experience CRF, and almost 25% of cancer survivors are also affected by CRF, lasting from months to years, even over a decade.^4,5 Therefore, it is essential to deliver an effective, safe and individualized therapeutic strategy for CRF.

Numerous pharmacological and non-pharmacological interventions have been evaluated in patients with cancer to target the fatigue outcome associated with the condition. However, there are no approved drugs used in CRF treatment. CAM interventions, including 5 categories (mind-body therapies, biologically based practices, manipulative and body-based practices, energy healing, and whole medical systems), are widely employed to improve human health among physicians, acupuncturists, and physical therapists domestically and abroad. In recent years, the effect of yoga, acupuncture, massage, and other CAM interventions on CRF has gained increasing interest. Many studies^6,7 report that CAM interventions can reduce fatigue in cancer patients and improve QoL, which is recommended by current guidelines as a kind of therapeutic strategy for the management of CRF.

However, current evidence shows heterogeneous effects, and related SRs have differences in the quality and outcomes, which causes the controversial efficacy of CAM on CRF. A feasible method to address this problem is to conduct an overview of previous SRs, which is exactly what we studied. This overview aimed to assess the quality of previous SRs and assimilate the existing evidence base to promote determining effective evidence-based interventions for CRF in adults and help clinicians to interpret the evidence.

Methods

The prospective protocol has been registered in PROSPERO with the number CRD42022304282. This overview of SRs was performed under the latest recommendations from the Cochrane Handbook and reported based on the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines (Supplemental Appendix 1). ⁸ No ethical approvals were required for this study.

Criteria for Inclusion and Exclusion

We only included SRs of RCTs and excluded SRs of non-randomized clinical trials, quasi-randomized clinical trials, and observational cohort studies. There were no restrictions on language and publication status. The target participants were adults greater than 18 years and patients with any type and stage of cancer. The interventions of SRs were CAM treatments, such as mind–body therapies (meditation, yoga, Tai Chi, etc.), manipulative and body-based practices (massage, reflexology, etc.), biologically based practices (dietary supplements, special diets, etc.), energy healing (reiki, therapeutic touch, etc.), or whole medical systems (traditional Chinese medicine, Ayurvedic medicine, etc.). The comparators were placebo, usual care, standard care, standard treatment, no interventions, or sham interventions. The primary outcome was fatigue, which must be reported in the included SRs. Secondary outcomes included QoL, depression, anxiety, sleep, pain, physical function, body mass index, biological indices, adherence, and adverse events.

Literature Searches

Two reviewers (QLZ and RZ) systematically and independently searched any published or ongoing SRs in the following databases from inception to September 2022: China National Knowledge Infrastructure (CNKI), Wan Fang database, Chinese biomedical literature service system (SinoMed), Chongqing VIP Database, MEDLINE (via PubMed), Web of Science, CINAHL, Scopus, OVID (including PsycInfo), Embase, PEDro, Cochrane Library, and PROSPERO. There were no limitations on language. Also, we manually searched all references of included studies to guarantee that no relevant SRs were eliminated by the electronic searches. We used the combination of text words and the Medical Subject Headings terms as a search strategy. Detailed electronic search strategies used in all databases are shown in Supplemental Appendix 2.

Study Selection

Two reviewers (QLZ and RZ) independently screened studies by titles and abstracts for possible inclusion. The full-text reports were also reviewed to assess appropriateness for final inclusion. Any inconsistencies were negotiated with a third reviewer (LXL) to resolve. Studies that were not eligible for enrollment were excluded, and the reasons were recorded.

Data Extraction

Two reviewers (YW and LXL) separately extracted essential information from included SRs into a normalized table after double-checking. Any disagreements were resolved by consensus with a third reviewer (QW). The extracted information included author, publication year, number of included primary studies and participants, type of cancer, type of CAM intervention and control, the approach of quality assessment, method of CRF evaluation, outcome indicators, and conclusion.

Assessment of Included Systematic Reviews

Two reviewers (QW and TTL) independently assessed the quality, risk of bias, integrity, and strength of evidence of included SRs. Disagreements and inconsistencies were resolved by a third reviewer (PJL).

AMSTAR-2, ⁹ a critical appraisal tool for SRs, was utilized to assess the quality of each SR. There are 16 items in the checklist, involving 7 critical items and 9 non-critical items. Responses for items were coded as “yes,” “no,” or “partial yes,” denoting a positive result, no relevant information provided, or partial adherence to the standard, respectively. We categorized the quality of included reviews as “high,” “moderate,” “low,” and “very low” based on the number of critical or non-critical weaknesses.

The ROBIS tool, ¹⁰ including 4 domains of 3 phases, was applied to assess the risk of bias of included reviews as “low,” “high,” or “unclear,” if the signaling questions for the domains were answered with “yes/probably yes,” “no/probably no,” or “no information,” respectively.

The PRISMA checklist, ⁸ including 27 items, was performed to evaluate the integrity of included reviews, and each item was responded to “yes,” “no,” or “partial yes.”

The GRADE system was used to assess the quality of evidence. ¹¹ RCTs were evaluated as “high quality” at first, and the levels lowered according to the presence of bias risk, publication bias, imprecision, inconsistency, or indirectness. The strength of evidence was categorized into 4 levels (very low, low, moderate, and high).

Data Analysis

A descriptive analysis of included SRs was conducted. We summarized and showed the pooled results of included SRs. For dichotomous data, the risk ratio (RR) or odds ratio (OR) with 95% confidence interval (CI) was recorded. For continuous data, the mean difference (MD) or standardized mean difference (SMD) with 95% CI was recorded. For SR without meta-analysis, the main results of primary RCTs were recorded.

Results

Literature Search

Of15 623 related studies retrieved through 13 databases and 10 records identified through manual searching, 9208 duplicates were excluded. We excluded 6266 studies after screening the titles and abstracts and 129 records after reviewing the full texts. Finally, we included 30 SRs for further analysis (Figure 1).^12-41 One hundred twenty-nine excluded records and the reasons for excluding them are provided in the Supplemental Appendix 3.

Figure 1.

Flowchart of the selection process.

Study Characteristics

Thirty SRs (27 meta-analyses) reported 5 to 67 RCTs and 394 to 5465 participants were eligible for inclusion. All SRs were published in the latest 15 years, 11 of which were published in 2021. Twenty SRs were conducted by China, while the rest of the 10 were performed by 8 other countries, including America, Brazil, Canada, etc. Nine SRs only included participants with breast cancer, one only enrolled patients with lung cancer, one did not report types of cancer, and others incorporated participants with mixed cancer. Three SRs did not conduct meta-analyses.^35,36,39 Among the 27 meta-analyses, 16 performed subgroup analyses, and 13 conducted sensitivity analyses.

Thirteen SRs assessed the effects of mind-body therapies (yoga, Tai Chi, qigong, music, and dance); 9 evaluated acupoint stimulation interventions (acupressure, acupuncture, moxibustion, and auricular point); 2 assessed manipulative and body-based practices (massage and reflexology); 3 evaluated dietary supplements; 2 assessed mixed CAM interventions; and 1 estimated kanglaite injection (a kind of traditional Chinese medicine injection).

Eighteen SRs reported on QoL; 11 involved depression and anxiety; 8 provided fatigue-related symptoms (sleep, pain, body mass index and physical function.); 1 reported the biological indices (cortisol and IL-6); 2 provided adherence; and 17 reported adverse events.

For the methodology quality assessment, most (n = 24) reviews used the Cochrane risk-of-bias tool, 2 used the Jadad scale,^37,39 4 assessed the quality with one each through the Delphi list, ¹⁸ PEDro scale, ²⁰ JBI’s tool, ²¹ and CONSORT guideline. ²³ The complete characteristics of included SRs are shown in Table 1 and Supplemental Appendix 4.

Table 1.

Characteristics of Included Systematic Reviews.

Author	Country	N/n	Cancer type	Intervention	Secondary outcomes
Liu et al 12	China	16/1133	Breast	Qigong, Tai Chi, yoga	No
Ma et al 13	China	9/703	Mixed	Qigong, Tai Chi	②⑩
Cheung et al 14	China	11/907	Mixed	Qigong	②④
Duan et al 15	China	15/1461	Mixed	Dance, qigong, Tai Chi, yoga	①②③④⑥⑦⑩
Zeng et al 16	China	13/592	Mixed	Qigong, Tai Chi	①②③⑥⑧
Qi et al 17	China	8/467	Mixed	Music	No
Buffart et al 18	Netherlands	16/783	Breast, lymphoma	Yoga	①②③④⑦⑩
O’Neill et al 19	Canada	24/1650	Breast	Yoga	①
Liu et al 20	New Zealand	16/1268	Breast	Tai Chi	①②④⑥
Song et al 21	China	16/1453	Mixed	Yoga	⑨
Ni et al 22	China	22/1410	Mixed	Tai Chi	④⑦⑧
Armer and Lutgendorf 23	America	29/1828	Mixed	Yoga	①②
Dong et al 24	China	17/2183	Breast	Yoga	⑨⑩
Tan et al 25	Australia	15/1302	Mixed	Acupressure, acupuncture	①⑩
He et al 26	China	7/804	Mixed	Acupuncture, moxibustion	①⑩
Zhang et al 27	China	33/2094	Breast	Acupuncture	①②③④⑤⑩
Yuanqing et al 28	China	20/1261	Breast	Acupuncture	①⑤⑦
Han et al 29	China	6/394	Mixed	Auricular point therapy	①⑩
Zhang et al 30	China	10/1327	Mixed	Acupuncture	⑩
Hsieh et al 31	China	14/776	Mixed	Acupressure	⑩
Ma et al 32	China	22/1628	Mixed	Moxibustion	①⑩
Yuan et al 33	China	11/832	Mixed	Acupuncture	①②③⑩
Pan et al 34	China	18/950	Breast	Massage	①②③⑤⑧
Wang et al 35	Australia	10/885	Breast	Massage, reflexology	①⑩
Pereira et al 36	Brazil	8/932	Breast	Dietary supplements	⑩
Najafi et al 37	Iran	5/683	NR	Ginseng	No
de Araujo et al 38	Brazil	7/427	Mixed	Guarana	⑩
Calcagni et al 39	France	41/3044	Mixed	Acupressure massage, reflexology	①②③④⑤
Tao et al 40	China	67/5465	Mixed	Acupoint stimulation, massage qigong, Tai Chi, TCM-FEMT	①②③④⑤
Liu et al 41	China	26/2209	Lung	Kanglaite injection	①⑤⑧⑩

①Quality of life; ②Depression; ③Anxiety; ④Sleep; ⑤Pain; ⑥Body mass index; ⑦Physical function; ⑧Biological indices; ⑨Adherence; ⑩Adverse events; N, number of randomized controlled trials; n, number of participants; NR, Not reported; TCM-FEMT, Traditional Chinese Medicine Five-Element Music Therapy.

The Methodological Quality of Included SRs

The methodological quality of all included SRs was no higher than “low” (Table 2). Twenty-nine SRs were rated as “critically low,” and only one ³¹ was rated as “low.” The questions with PICO were described in all studies except three,^34,36,37 and prior study protocols were established in 8 studies. Only one SR ¹³ explained why selected RCTs for inclusion. All except one SR, ⁴¹ which did not provide any search terms or strategies, undertook a comprehensive literature search. Five SRs did not select studies or extract data in duplicates. One SR ¹² did not describe comparators of included studies. Two SRs^37,39 did not use proper techniques to assess the risk of bias. No SRs afforded checklists of excluded studies and the reasons for excluding them. Only one SR ³⁸ reported the funding sources for the primary studies. Most SRs reported potential conflicts of interest, but few assessed publication bias. All meta-analyses used appropriate methods for results combination, but only one ³¹ conducted a meta-regression analysis to evaluate possible impacts of bias risk on the results. When discussing the results, 26 SRs (86.67%) accounted for the bias risk, and 25 SRs (83.33%) considered heterogeneity.

Table 2.

AMSTAR-2 Tool of Methodological Quality Assessment of Included Systematic Reviews.

Study	I1	I2*	I3	I4*	I5	I6	I7*	I8	I9*	I10	I11*	I12	I13*	I14	I15 *	I16	Quality
Liu et al 12	Y	Y	N	PY	Y	Y	N	N	Y	N	Y	N	N	N	N	Y	Critically low
Ma et al 13	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	Y	Y	N	Y	Critically low
Cheung et al 14	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	Y	Y	N	Y	Critically low
Duan et al 15	Y	N	N	PY	Y	Y	N	PY	Y	N	Y	N	Y	N	Y	Y	Critically low
Zeng et al 16	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	Y	N	N	Y	Critically low
Qi et al 17	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	Y	Y	Y	Y	Critically low
Buffart et al 18	Y	N	Y	PY	Y	N	N	Y	Y	N	Y	N	Y	Y	N	Y	Critically low
O’Neill et al 19	Y	N	N	PY	Y	Y	N	PY	Y	N	Y	N	Y	Y	N	Y	Critically low
Liu et al 20	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	Y	Y	N	Y	Critically low
Song et al 21	Y	Y	N	PY	Y	Y	N	PY	Y	N	Y	N	Y	Y	N	N	Critically low
Ni et al 22	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	Y	Y	N	Y	Critically low
Armer and Lutgendorf 23	Y	N	N	PY	Y	N	N	Y	Y	N	Y	N	Y	Y	Y	Y	Critically low
Dong et al 24	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	N	Y	Y	Y	Critically low
Tan et al 25	Y	PY	N	PY	Y	Y	N	PY	Y	N	Y	N	Y	Y	N	Y	Critically low
He et al 26	Y	Y	N	PY	Y	Y	N	PY	Y	N	Y	N	Y	Y	N	Y	Critically low
Zhang et al 27	Y	Y	N	PY	Y	Y	N	Y	Y	N	Y	N	Y	Y	N	Y	Critically low
Yuanqing et al 28	Y	N	N	PY	Y	Y	N	PY	Y	N	Y	N	Y	Y	N	Y	Critically low
Han et al 29	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	Y	Y	Y	Y	Critically low
Zhang et al 30	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	N	N	N	Y	Critically low
Hsieh et al 31	Y	PY	N	PY	Y	Y	N	Y	Y	N	Y	Y	Y	Y	Y	Y	Low
Ma et al 32	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	N	Y	Y	N	Y	Critically low
Yuan et al 33	Y	N	N	PY	Y	Y	N	PY	Y	N	Y	N	Y	Y	Y	N	Critically low
Pan et al 34	N	N	N	PY	N	N	N	Y	Y	N	Y	N	Y	Y	N	N	Critically low
Wang et al 35	Y	N	N	PY	Y	Y	N	Y	Y	N	no meta	no meta	Y	Y	no meta	Y	Critically low
Pereira et al 36	N	PY	N	PY	Y	N	N	Y	Y	N	no meta	no meta	Y	Y	no meta	Y	Critically low
Najafi et al 37	N	N	N	PY	Y	Y	N	Y	N	N	Y	N	N	Y	N	Y	Critically low
de Araujo et al 38	Y	Y	N	PY	Y	Y	N	Y	Y	Y	Y	N	Y	Y	N	Y	Critically low
Calcagni et al 39	Y	N	N	PY	N	Y	N	PY	N	N	no meta	no meta	Y	Y	no meta	Y	Critically low
Tao et al 40	Y	N	N	PY	Y	Y	N	PY	Y	N	Y	N	Y	Y	N	Y	Critically low
Liu et al 41	Y	N	N	N	Y	Y	N	PY	Y	N	Y	N	Y	N	N	N	Critically low

*

The key items of the AMSTAR-2 tool; I, Item; Y, Yes; N, No; PY, Partial yes.

Risk of Bias of Included SRs

We applied the ROBIS tool to evaluate the risk of bias in included SRs (Table 3, Figure 2). In phase 1, the target questions of all SRs matched or partially matched the study questions. In phase 2, we identified the risk of bias in the study process through 4 domains: study eligibility (domain 1), study selection (domain 2), data collection and quality appraisal (domain 3), and data synthesis and findings (domain 4). There were 22, 16, 17, and 8 SRs with low risk of bias in domains 1, 2, 3, and 4, respectively. One SRs were assessed as unclear risk of bias in domains 2 and 3. We considered the overall bias risk of included SRs in phase 3. Nineteen SRs were demonstrated as low bias risk, and 11 were evaluated as high bias risk.

Table 3.

ROBIS Tool of Risk of Bias of Included Systematic Reviews.

Study	Phase 1	Phase 2				Phase 3
	Assessing relevance	Study eligibility criteria	Identification and selection of studies	Data collection and study appraisal	Synthesis and findings	Risk of bias in the review
Liu et al 12	PY
Ma et al 13	Y
Cheung et al 14	PY
Duan et al 15	PY
Zeng et al 16	PY
Qi et al 17	Y
Buffart et al 18	Y
O’Neill et al 19	Y
Liu et al et al 20	PY
Song et al 21	Y
Ni et al 22	PY
Armer and Lutgendorf 23	Y
Dong et al 24	Y
Tan et al 25	Y
He et al 26	Y
Zhang et al 27	PY
Yuanqing et al 28	PY
Han et al 29	Y
Zhang et al 30	Y
Hsieh et al 31	Y
Ma et al 32	PY
Yuan et al 33	Y
Pan et al 34	PY
Wang et al 35	Y
Pereira et al 36	PY
Najafi et al 37	PY
de Araujo et al 38	Y
Calcagni et al 39	Y
Tao et al 40	PY
Liu et al 41	PY

Low risk; High risk; Unclear risk; Y, Yes; PY, Partial yes.

Figure 2.

Graphical presentation for ROBIS tool of risk of bias.

Reporting Quality of Included SRs

Reporting quality and integrity of included SRs were evaluated with the PRISMA checklist, including 27 items (Table 4). No SRs reported all items. Ten SRs sufficiently reported over 70% of all items, and only one reported less than 50%. All SRs adequately introduced their rationales and objectives. In the methods section, eligibility criteria and information sources were well reported (100%). Most SRs specified how to identify studies, extract data, and evaluate the quality. In the results section, more than 85% of SRs provided general characteristics, bias risk, and primary outcomes of included studies. 86.7% of SRs discussed any limitations and implications of included studies. Ninety percent of SRs described the funding sources and declared any competing interests. Overall, the central reporting limitations consisted of registration and protocol (23.3%), abstract (0%), search strategy (10%), assessment and results of reporting bias (26.7%), assessment and results of evidence certainty (10%, 13.3%), and study selection (3.3%).

Table 4.

Compliance of PRISMA Checklist of Included Systematic Reviews.

Items	Liu et al 12	Ma et al 13	Cheung et al 14	Duan et al 15	Zeng et al 16	Qi et al 17	Buffart et al 18	O’Neill et al 19	Liu et al 20	Song et al 21	Ni et al 22	Armer and Lutgendorf 23	Dong et al 24	Tan et al 25	He et al 26	Zhang et al 27	Yuanqing et al 28	Han et al 29	Zhang et al 30	Hsieh et al 31	Ma et al 32	Yuan et al 33	Pan et al 34	Wang et al 35	Pereira et al 36	Najafi et al 37	de Araujo et al 38	Calcagni et al 39	Tao et al 40	Liu et al 41	Compliance (%)
1	N	N	Y	Y	Y	N	Y	Y	Y	Y	Y	N	N	N	Y	Y	Y	Y	N	Y	Y	N	Y	Y	Y	Y	Y	Y	N	Y	70.0
2	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	0.0
3	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100.0
4	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100.0
5	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100.0
6	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100.0
7	N	N	N	N	N	N	N	Y	N	N	N	N	N	N	N	Y	N	Y	N	N	N	N	N	N	N	N	N	N	N	N	10.0
8	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	N	Y	93.3
9	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	N	Y	Y	Y	Y	Y	93.3
10	Y	Y	Y	Y	N	N	N	Y	Y	Y	N	Y	N	Y	Y	Y	N	Y	N	Y	Y	Y	N	Y	Y	Y	N	Y	Y	Y	70.0
11	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	96.7
12	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	N	N	Y	N	Y	Y	Y	83.3
13	N	N	N	Y	Y	Y	Y	N	N	Y	Y	N	Y	Y	N	N	N	Y	N	N	Y	Y	N	Y	Y	N	N	Y	N	Y	50.0
14	N	N	Y	N	N	Y	N	Y	N	Y	N	Y	Y	N	N	N	N	N	N	Y	N	Y	N	N	N	N	N	N	N	N	26.7
15	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	Y	N	N	N	N	N	N	N	Y	N	Y	N	N	N	10.0
16	N	N	N	N	N	N	N	Y	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	N	3.3
17	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	96.7
18	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	93.3
19	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	N	Y	Y	Y	N	Y	Y	Y	Y	Y	86.7
20	Y	N	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	93.3
21	N	N	Y	Y	N	Y	N	N	N	Y	N	Y	Y	N	N	N	N	N	N	Y	N	Y	N	N	N	N	N	N	N	N	26.7
22	N	N	N	N	N	N	N	N	N	N	Y	N	N	N	N	N	Y	N	N	N	N	N	N	N	Y	N	Y	N	N	N	13.3
23	N	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	86.7
24	Y	N	N	N	N	N	N	N	N	Y	N	N	N	Y	N	Y	N	N	N	Y	N	N	N	N	Y	N	Y	N	N	N	23.3
25	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	N	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	90.0
26	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	N	90.0
27	Y	N	Y	Y	N	Y	N	Y	N	N	Y	N	N	Y	N	Y	N	Y	Y	Y	N	N	N	Y	N	N	Y	Y	N	N	46.7
Compliance (%)	63.0	51.9	74.1	74.1	59.3	70.4	63.0	77.8	63.0	66.7	66.7	63.0	66.7	70.4	51.9	70.4	66.7	70.4	55.6	74.1	63.0	66.7	48.1	66.7	66.7	59.3	70.4	70.4	55.6	63.0	-

Abbreviations: Y, Yes; N, No.

Evidence Quality of Included SRs

Given that 3 SRs did not combine the results for meta-analysis, we only used the GRADE system to evaluate the strength of evidence for CRF outcomes (51 pooled results) and other significant secondary outcomes (69 pooled results) reported in the other 27 SRs. Among 51 pooled results of CRF, 2 were assessed as “high” quality, 21 as “moderate” quality, 18 as “low” quality, and 10 as “very low” quality. As for other secondary outcomes (QoL, depression, anxiety, etc.), 3 of 69 pooled results were assessed as “high” quality, 30 as “moderate” quality, 21 as “low” quality, and 15 as “very low” quality. The certainty of the evidence was downgraded mainly due to imprecision and inconsistency. Tables 5 and 6 show the reasons for downgrading.

Table 5.

Summary and GRADE Quality of Evidence for CRF Outcomes Reported in the Included Systematic Reviews.

Study	Intervention	N (n)	Effect size	GRADE SR of RCTs = +4					GRADE quality
				Risk of bias	Publication bias	Imprecision	Inconsistency	Indirectness
Liu et al 12	Mind-body therapy vs control	17 (1133)	SMD = 0.59, 95% CI (0.27, 0.92), P = .0004	0	0	0	−1④	0	⊕⊕⊕○ Moderate
	Mind-body therapy vs control (excluding one study)	16 (1026)	SMD = 0.43, 95% CI (0.22, 0.64), P < .00001	0	0	0	0	0	⊕⊕⊕⊕High
	Tai Chi vs control	2 (74)	SMD = 0.96, 95% CI (0.10, 1.82), P = .03	0	0	−1③	0	0	⊕⊕⊕○Moderate
	Yoga vs control	13 (912)	SMD = 0.59, 95% CI (0.18, 0.99), P < .0001	0	0	0	−1④	0	⊕⊕⊕○Moderate
	Other types vs control	2 (147)	SMD = 0.32, 95% CI (−0.21, 0.84), P = .23	0	0	−1③	0	0	⊕⊕⊕○Moderate
Ma et al 13	Tai Chi/qigong vs control (scores at ≤ 12-week follow-up)	3 (201)	SMD = −0.00, 95% CI (−0.42, 0.41), P = .99	0	0	−1③	0	0	⊕⊕⊕○Moderate
	Tai Chi/qigong vs control (scores at > 12-week follow-up)	2 (168)	SMD = −0.05, 95% CI (−0.26, −0.35), P = .77	0	0	−1③	0	0	⊕⊕⊕○Moderate
	Tai Chi/qigong vs control (change scores at ≤12-week follow-up)	2 (137)	SMD = 1.03, 95% CI (0.61, 1.44), P < .00001	0	−1②	0	0	0	⊕⊕⊕○Moderate
Cheung et al 14	Qigong vs. control (post intervention)	8 (729)	SMD = -0.89, 95% CI (-1.59, -0.19)	−1①	0	0	−2⑤	0	⊕○○○Very low
	Qigong vs. control (3-month follow-up)	4 (363)	SMD = -0.14, 95% CI (-0.64, 0.36)	−1①	0	0	−1④	0	⊕⊕○○Low
Duan et al 15	Mind-body therapy vs control	10 (694)	SMD = −0.47, 95% CI (−0.88, −0.06), P = .02	−1①	0	0	−1④	0	⊕⊕○○Low
	Qigong vs control	1 (96)	SMD = 0.19, 95% CI (−0.21, 0.59), P = .34	-	-	-	-	-	-
	Tai Chi vs control	3 (217)	SMD = −0.95, 95% CI (−1.48, −0.43), P = 0.0004	0	−1②	0	0	0	⊕⊕⊕○Moderate
	Yoga vs control	6 (381)	SMD = −0.33, 95% CI (−0.87, 0.22), P = .24	−1①	0	0	−1④	0	⊕⊕○○Low
Zeng et al 16	Qigong vs control (change scores at 12-week follow-up)	2 (257)	SMD = −0.93, 95% CI (−1.80, −0.06), P = .04	0	−1②	−1③	−2⑤	0	⊕○○○Very low
Qi et al 17	Music vs control	8 (467)	SMD = −0.88, 95% CI (−1.49, −0.26), P = .005	−1①	0	0	−1④	0	⊕⊕○○Low
Buffart et al 18	Yoga vs control	7 (382)	SMD = −0.51, 95% CI (−0.79, −0.22), P = .001	−1①	0	0	0	0	⊕⊕⊕○Moderate
O’Neill et al 19	Yoga vs non-active comparators	18 (1066)	SMD = −0.30, 95% CI (−0.51, −0.08)	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Yoga vs active comparators	6 (NR)	SMD = −0.17, 95% CI (−0.50, 0.17)	−1①	0	−1③	0	0	⊕⊕○○Low
	Yoga vs control (low risk of bias)	14 (839)	SMD = −0.20, 95% CI (−0.34, −0.06)	0	0	0	0	0	⊕⊕⊕⊕High
Liu 20	Tai Chi vs conventional supportive care (3-mo follow-up)	2 (174)	SMD = −0.46, 95% CI (−1.09, 0.17), P = .15	0	−1②	−1③	0	0	⊕⊕○○Low
	Tai Chi vs conventional supportive care (6-mo follow-up)	2 (163)	SMD = −0.16, 95% CI (−0.98, 0.67), P = .71	0	−1②	0	0	0	⊕⊕⊕○Moderate
	Tai Chi plus conventional therapy vs conventional therapy (3-month follow-up)	2 (116)	SMD = −0.91, 95% CI (−1.30, −0.53), P < .0001	0	−1②	−1③	0	0	⊕⊕○○Low
Song et al 21	Yoga vs routine care	11 (1293)	SMD = −0.52, 95% CI (–0.86, –0.18), P = .003	0	0	0	−1④	0	⊕⊕⊕○Moderate
Ni et al 22	Tai Chi vs control	3 (234)	SMD = −0.04, 95% CI (−0.30, 0.22), P = .76	0	0	0	−1④	0	⊕⊕⊕○Moderate
Armer and Lutgendorf 23	Yoga vs control	29 (1828)	SMD = 0.45, 95% CI (0.09, 0.82), P = .013	0	−1②	0	−2⑤	0	⊕○○○Very low
Dong et al 24	Yoga vs control	17 (2183)	SMD = −0.31, 95% CI (−0.52, −0.10), P = .003	0	−1②	0	−1④	0	⊕⊕○○Low
Tan et al 25	Acupoint stimulation vs standard treatment/care	9 (451)	SMD = −0.95, 95% CI (−1.66, −0.25), P = .008	0	−1②	0	−1④	0	⊕⊕○○Low
	Acupoint stimulation vs sham acupoint stimulation	4 (224)	SMD = −0.28, 95% CI (−0.54, −0.01), P = .04	0	−1②	0	0	0	⊕⊕⊕○Moderate
	Acupuncture vs standard treatment/care	7 (361)	SMD = −1.25, 95% CI (−2.05, −0.45), P = .002	0	−1②	0	−2⑤	0	⊕○○○Very low
	Acupuncture vs standard treatment/care (excluding one study)	6 (281)	SMD = −0.74, 95% CI (−0.99, −0.50), P < .00001	0	−1②	0	0	0	⊕⊕⊕○Moderate
	Acupuncture vs sham acupuncture	2 (123)	SMD = −0.29, 95% CI (−0.65, 0.07), P = .11	0	−1②	0	0	0	⊕⊕⊕○Moderate
	Acupressure vs sham acupressure	2 (100)	SMD = −0.26, 95% CI (−0.66, 0.14), P = .20	−1①	−1②	0	0	0	⊕⊕○○Low
He et al 26	Moxibustion plus routine care vs routine care (PFS, score 0-3 points)	2 (198)	OR = 2.96, 95% CI (1.00, 8.73), P = .05	−1①	0	−1③	0	0	⊕⊕○○Low
	Moxibustion plus routine care vs routine care (PFS, score 4-6 points)	2 (198)	OR = 0.90, 95% CI (0.12, 6.75), P = .92	−1①	0	−1③	−2⑤	0	⊕○○○Very low
	Moxibustion plus routine care vs routine care (PFS, score ≥7 points)	2 (198)	OR = 0.16, 95% CI (0.07, 0.37), P < .0001	−1①	0	−1③	0	0	⊕⊕○○Low
	Moxibustion plus routine care vs routine care	1 (200)	OR = −1.01, 95% CI (−1.31, −0.72), P < .00001	-	-	-	-	-	-
Zhang et al 27	Acupuncture vs control	6 (590)	SMD = −0.39, 95% CI (−0.55, −0.22), P < .00001	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Acupuncture vs control (high-quality studies)	3 (136)	SMD = −0.34, 95% CI (−0.68, −0.00), P = .05	0	0	−1③	0	0	⊕⊕⊕○Moderate
	Acupuncture vs sham acupuncture	4 (250)	SMD = −0.27, 95% CI (−0.52, −0.02), P < .04	0	0	−1③	0	0	⊕⊕⊕○Moderate
	Acupuncture vs no intervention	4 (275)	SMD = −0.50, 95% CI (−0.76, −0.25), P < .0001	−1①	0	−1③	0	0	⊕⊕○○Low
Pan et al 28	Acupuncture vs control	4 (177)	SMD = −1.19, 95% CI (−2.25, −0.12), P = .02	0	−1②	−1③	−1④	0	⊕○○○Very low
Han et al 29	Auricular point therapy plus standard care vs standard care	5 (363)	RR = 1.76, 95% CI (1.42, 2.17), P < .0001	−1①	−1②	−1③	0	0	⊕○○○Very low
Zhang et al 30	Acupuncture vs control	10 (1327)	SMD = −1.26, 95% CI (−1.80, −0.71), P < .00001	0	0	0	−2⑤	0	⊕⊕○○Low
Hsieh et al 31	Acupressure vs control	14 (776)	SMD = −0.87, 95% CI (−1.19, −0.55), P = .000	0	0	0	−1④	0	⊕⊕⊕○Moderate
Ma et al 32	Moxibustion vs control	10 (780)	SMD = −1.17, 95% CI (−1.44, −0.90), P < .00001	0	−1②	0	0	0	⊕⊕⊕○Moderate
Yuan et al 33	Acupuncture vs control	11 (832)	SMD = −1.06, 95% CI (−1.73, −0.40), P = .002	0	0	0	−2⑤	0	⊕⊕○○Low
Pan et al 34	Massage vs control	5 (667)	SMD = −0.61, 95% CI (−1.09, −0.13), P = .01	−1①	0	0	−1④	0	⊕⊕○○Low
Wang et al 35	Reflexology vs routine treatment and care	1 (57)	P ≤ .01	-	-	-	-	-	-
	Massage vs routine treatment and care	1 (98)	P < .05	-	-	-	-	-	-
	Massage vs light touch/wait list control	1 (56)	P < .0001	-	-	-	-	-	-
	Reflexology vs routine treatment and care	1 (60)	P < .05	-	-	-	-	-	-
	Reflexology vs routine treatment and care	1 (207)	P < .01 (2 and 3-week follow-up), P = .15 (4-week follow-up)	-	-	-	-	-	-
	Massage vs routine treatment and care	1 (20)	P < .001	-	-	-	-	-	-
	Reflexology vs light touch/routine treatment and care	1 (243)	P < .01 (vs routine treatment and care), P = .02 (versus light touch)	-	-	-	-	-	-
	Massage vs routine treatment and care	1 (43)	a smaller increase in fatigue (3.6 ± 2.8vs 2.5 ± 1.5)	-	-	-	-	-	-
	Massage vs routine treatment and care	1 (29)	P = .056 (post intervention), P < .01 (11-week follow-up)	-	-	-	-	-	-
	Massage vs routine treatment and care	1 (72)	P = .06 (post intervention), P = .048 (11-week follow-up)	-	-	-	-	-	-
Pereira et al 36	Guarana extract vs crystalline microcellulose	1 (39)	P < .05 (the total fatigue and the dimensions of fatigue)	-	-	-	-	-	-
	Guarana extract vs control	1 (45)	P < .01 (21 and 49-day follow-up)	-	-	-	-	-	-
	Acetyl-L-carnitine vs control	1 (395)	P > .05	-	-	-	-	-	-
	Branched-chain amino acids, CoQ10, and acetyl-L-carnitine vs control	1 (57)	P < .05 (the worst level of fatigue and current feeling of fatigue)	-	-	-	-	-	-
	Coconut oil vs control	1 (60)	P > .05	-	-	-	-	-	-
	CoQ10 plus vitamin E vs control	1 (139)	P = .965 (24-week follow-up)	-	-	-	-	-	-
	Chlorella granules vs control	1 (27)	P > .05	-	-	-	-	-	-
	Chlorella extract vs control	1 (27)	P < .01	-	-	-	-	-	-
	Fatigue reduction diet vs health care control	1 (29)	P = .01 (3-month follow-up)	-	-	-	-	-	-
Najafi et al 37	Ginseng vs control	5 (683)	SMD = 1.24, 95% CI (0.05, 2.42), P = .000	0	0	0	−2⑤	0	⊕⊕○○Low
de et al 38	Guarana vs placebo	3 (139)	SMD = −0.02, 95% CI (−1.54, 1.50), P = .98	0	0	−1③	−2⑤	0	⊕○○○Very low
Calcagni et al 39	Massage vs. usual care plus attention	1 (20)	P < 0.05	-	-	-	-	-	-
	Massage vs. light touch/wait list control	1 (56)	P < 0.05 (6-week follow-up)	-	-	-	-	-	-
	Reflexology vs. standard care/lay foot manipulation	1 (286)	P > 0.05	-	-	-	-	-	-
	Reflexology vs. standard care	1 (60)	P < 0.05 (3, 4, and 5-week follow-up	-	-	-	-	-	-
	Acupressure vs. sham acupressure	1 (48)	P < 0.05 (6-week follow-up), P < 0.01 (12-week follow-up)	-	-	-	-	-	-
Tao et al 40	Acupuncture vs control (PFS)	2 (266)	OR = 2.40, 95% CI (1.40, 4.13), P = .002	−1①	0	−1③	0	0	⊕⊕○○Low
	Acupuncture vs control (BFI)	5 (366)	SMD = −1.41, 95% CI (−2.58, −0.24), P = 0.02	0	0	−1③	−2⑤	0	⊕○○○Very low
	Acupuncture/Tai Chi/qigong vs control (BFI)	7 (560)	SMD = −1.19, 95% CI (−2.19, −0.18), P < .05	-	-	-	-	-	-
Liu et al 41	Kanglaite injection plus chemotherapy vs chemotherapy	2 (202)	RR = 0.75, 95% CI (0.04, 15.32)	−1①	0	−1③	−1④	0	⊕○○○Very low

①Serious risk of bias; ②Suspicion of publication bias; ③The cumulative sample size was not enough, and/or 95% confidence interval was very wide; ④I²-value of the results was large ( ≥70%); ⑤I²-value of the results was very large ( ≥90%); BFI, Brief Fatigue Inventory; NR, Not reported; PFS, Piper Fatigue Scale.

Table 6.

GRADE Quality of Evidence for Significant Secondary Outcomes Reported in the Included Systematic Reviews.

Study	Intervention	N (n)	Effect size	GRADE SR of RCTs = +4					Quality of evidence
				Risk of bias	Publication bias	Imprecision	Inconsistency	Indirectness
Quality of life
Buffart et al 18	Yoga vs control: overall QoL	5 (291)	SMD = 0.37, 95% CI (0.11, 0.62), P = .004	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Yoga vs control: emotional function	5 (402)	SMD = 0.49, 95% CI (0.16, 0.81), P = .003	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Yoga vs control: social function	6 (440)	SMD = 0.33, 95% CI (0.12, 0.54), P = 002	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Yoga vs control: functional well-being	4 (253)	SMD = 0.31, 95% CI (0.04, 0.58), P = .03	−1①	0	0	0	0	⊕⊕⊕○Moderate
Han et al 29	Auricular point plus standard care vs standard care: overall QoL	3 (215)	MD = 7.34, 95% CI (5.11, 9.57), P < .00001	0	−1②	−1③	0	0	⊕⊕○○Low
	Auricular point plus standard care vs standard care: role function	3 (215)	MD = 8.99, 95% CI (7.45, 10.53), P < .00001	0	−1②	−1③	0	0	⊕⊕○○Low
	Auricular point plus standard care vs standard care: emotional function	3 (215)	MD = 8.75, 95% CI (6.51, 10.98), P < .00001	0	−1②	−1③	0	0	⊕⊕○○Low
	Auricular point plus standard care vs standard care: cognitive function	3 (215)	MD = 7.17, 95% CI (4.85, 9.50), P < .00001	0	−1②	−1③	0	0	⊕⊕○○Low
	Auricular point plus standard care vs standard care: social function	3 (215)	MD = 5.64, 95% CI (4.17, 7.10), P < .00001	0	−1②	−1③	0	0	⊕⊕○○Low
He et al 26	Moxibustion plus routine care vs routine care: global QoL	2 (278)	SMD = 0.81, 95% CI (0.20, 1.43), P = .01	0	−1②	0	−1④	0	⊕⊕○○Low
	Moxibustion plus routine care vs routine care: emotional function	2 (278)	SMD = 0.97, 95% CI (0.72, 1.22), P < .00001	0	−1②	0	0	0	⊕⊕⊕○Moderate
Liu et al 41	Kanglaite injection plus chemotherapy vs chemotherapy	18 (1590)	RR = 2.05, 95% CI (1.60, 2.64), P < .001	−1①	0	0	−1④	0	⊕⊕○○Low
Liu et al 20	Tai Chi vs conventional supportive care (3-month follow-up)	2 (952)	SMD = 0.32, 95% CI (0.07, 0.56), P = .01	0	−1②	0	0	0	⊕⊕⊕○Moderate
	Tai Chi plus conventional therapy vs conventional therapy (3-month follow-up)	4 (2382)	SMD = 0.34, 95% CI (0.26, 0.43), P < .00001	0	−1②	0	0	0	⊕⊕⊕○Moderate
	Tai Chi plus conventional therapy vs conventional therapy (6-month follow-up)	4 (2382)	SMD = 0.40, 95% CI (0.31, 0.49), P < .00001	0	−1②	0	0	0	⊕⊕⊕○Moderate
Ma et al 32	Moxibustion vs control: global QoL	9 (471)	SMD = 0.88, 95% CI (0.54, 1.22), P < .00001	0	0	0	0	0	⊕⊕⊕⊕High
	Moxibustion vs control: role function	6 (322)	MD = 9.39, 95% CI (2.22, 16.56), P = .01	0	0	−1③	−1④	0	⊕⊕○○Low
	Moxibustion vs control: emotional function	8 (447)	MD = 11.71, 95% CI (4.41, 19.01), P = .002	0	0	−1③	−1④	0	⊕⊕○○Low
	Moxibustion vs control: cognitive function	7 (382)	MD = 7.96, 95% CI (3.08, 12.85), P = .001	0	0	−1③	0	0	⊕⊕⊕○Moderate
	Moxibustion vs control: social function	7 (382)	MD = 11.01, 95% CI (7.46, 14.56), P < .00001	0	0	−1③	0	0	⊕⊕⊕○Moderate
Ni et al 22	Tai Chi vs control: physical of QoL	9 (679)	SMD = 0.34, 95% CI (0.09, 0.59), P = .007	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Tai Chi vs control: psychological of QoL	9 (681)	SMD = 0.60, 95% CI (0.12, 1.08), P = .01	−1①	0	0	−1④	0	⊕⊕○○Low
O’Neill et al 19	Yoga vs non-active comparators	10 (568)	SMD = 0.27, 95% CI (0.07, 0.46)	−1①	0	0	0	0	⊕⊕⊕○Moderate
Tao et al 40	Acupuncture vs control: QoL (means of KPS)	4 (450)	MD = 4.55, 95% CI (0.75, 8.34), P = .02	−1①	0	−1③	−1④	0	⊕○○○Very low
	Acupuncture vs control: QoL (remission rate of KPS)	3 (242)	OR = 2.32, 95% CI (1.26, 4.27), P = .007	−1①	0	−1③	0	0	⊕⊕○○Low
	Acupoint stimulation/massage/qigong/Tai Chi/TCM-FEMT vs control: QoL (cancer-specific QOL scales)	9 (686)	SMD = 0.97, 95% CI (0.18, 1.76), P = .02	−1①	0	0	−2⑤	0	⊕○○○Very low
	Massage vs control: QoL (cancer-specific QOL scales)	2 (194)	SMD = 1.31, 95% CI (0.21, 2.41), P = .02	−1①	0	−1③	−2⑤	0	⊕○○○Very low
Zeng et al 16	Qigong vs control (change scores at 12-wk follow-up)	4 (395)	MD = 6.57, 95% CI (2.32, 10.84), P = .002	−1①	0	−1③	−2⑤	0	⊕○○○Very low
Zhang et al 27	Acupuncture vs control	2 (154)	MD = 10.09, 95% CI (7.26, 12.92), P < .0001	0	−1②	−1③	0	0	⊕⊕○○Low
Depression
Armer and Lutgendorf 23	Yoga vs control	12 (587)	SMD = 0.72, 95% CI (0.20, 1.24), P = .007	0	0	0	−1④	0	⊕⊕⊕○Moderate
Buffart et al 18	Yoga vs control	6 (413)	SMD = −0.69, 95% CI (−1.02, −0.37), P < .001	−1①	0	0	0	0	⊕⊕⊕○Moderate
Duan et al 15	Mind-body therapy vs control	8 (518)	SMD = −0.21, 95% CI (−0.42, −0.01), P = .04	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Yoga vs control	5 (206)	SMD = −0.33, 95% CI (−0.60, −0.05), P = .02	−1①	0	0	0	0	⊕⊕⊕○Moderate
Tao et al 40	Acupuncture/massage/TCM-FEMT vs control	7 (826)	MD = −7.43, 95% CI (−12.96, −1.91), P = .008	−1①	0	−1③	−2⑤	0	⊕○○○Very low
	TCM-FEMT vs control	2 (128)	SMD = −2.96, 95% CI (−5.60, −0.32), P = .03	−1①	0	−1③	−2⑤	0	⊕○○○Very low
Anxiety
Buffart et al 18	Yoga vs control	6 (455)	SMD = −0.77, 95% CI (−1.08, −0.46), P < .001	−1①	0	0	0	0	⊕⊕⊕○Moderate
Duan et al 15	Mind-body therapy vs control	6 (326)	SMD = 0.27, 95% CI (0.01, 0.54), P = 04	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Yoga vs control	2 (149)	SMD = 0.52, 95% CI (0.19, 0.86), P = .002	−1①	0	0	0	0	⊕⊕⊕○Moderate
Tao et al 40	Acupuncture/massage/TCM-FEMT vs control	4 (624)	MD = −7.82, 95% CI (−13.74, −1.90), P = .010	−1①	0	−1③	−2⑤	0	⊕○○○Very low
Zhang et al 27	Acupuncture vs control	3 (161)	MD = −0.37, 95% CI (−0.68, −0.05), P = .02	0	−1②	0	0	0	⊕⊕⊕○Moderate
Sleep
Cheung et al 14	Qigong vs control (post intervention)	10 (812)	SMD = −1.28, 95% CI (−2.01, −0.55)	0	0	−1③	−2⑤	0	⊕○○○Very low
Duan et al 15	Mind-body therapy vs control	8 (834)	SMD = −0.66, 95% CI (−0.71, −0.60), P < .00001	0	−1②	0	0	0	⊕⊕⊕○Moderate
	Yoga vs control	5 (565)	SMD = −0.66, 95% CI (−0.72, −0.60), P < .00001	−1①	−1②	0	0	0	⊕⊕○○Low
Ni et al 22	Tai Chi vs control (excluding one study)	2 (128)	SMD = −0.37, 95% CI (−0.72, −0.02), P = .04	−1①	−1②	0	0	0	⊕⊕○○Low
Tao et al 40	Acupuncture/massage/qigong vs control	4 (646)	SMD = −2.53, 95% CI (−4.17, −0.89), P = .003	−1①	0	−1③	−2⑤	0	⊕○○○Very low
	Acupuncture vs control	2 (151)	SMD = −1.66, 95% CI (−2.69, −0.63), P = .002	−1①	0	−1③	−1④	0	⊕○○○Very low
Zhang et al 27	Acupuncture vs control	5 (342)	SMD = −0.50, 95% CI (−0.71, −0.28), P < .00001	0	0	0	0	0	⊕⊕⊕⊕High
Pain
Liu et al 41	Kanglaite injection plus chemotherapy vs chemotherapy	4 (377)	RR = 2.90, 95% CI (2.11, 3.98)	0	0	0	−1④	0	⊕⊕⊕○Moderate
Ma et al 32	Moxibustion vs control	5 (242)	SMD = −0.70, 95% CI (−1.37, −0.03), P = .04	0	0	−1③	−1④	0	⊕⊕○○Low
Tao et al 40	Acupuncture vs control	8 (591)	SMD = −0.30, 95% CI (−0.56, −0.03), P = .03	−1①	0	0	0	0	⊕⊕⊕○Moderate
Zhang et al 27	Acupuncture vs control: pain	3 (160)	SMD = −0.83, 95% CI (−1.16, −0.51), P < .00001	−1①	−1②	0	0	0	⊕⊕○○Low
	Acupuncture vs control: worst pain	4 (442)	MD = −1.60, 95% CI (−2.63, −0.57), P = .003	0	0	−1③	−1④	0	⊕⊕○○Low
	Acupuncture vs control: average pain	3 (376)	MD = −0.86, 95% CI (−1.32, −0.4), P = .0003	−1①	−1②	0	0	0	⊕⊕○○Low
	Acupuncture vs control: pain interference	6 (508)	MD = −1.04, 95% CI (−1.66, −0.42), P = .001	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Acupuncture vs control: pain severity	6 (508)	MD = −1.28, 95% CI (−2.17, −0.39), P = .005	−1①	0	−1③	−1④	0	⊕○○○Very low
Body mass index
Duan et al 15	Mind-body therapy vs control	2 (108)	SMD = 1.31, 95% CI (0.04, 2.58), P = .04	0	−1②	−1③	0	0	⊕⊕○○Low
Tao et al 40	Tai Chi vs control	4 (245)	SMD = −0.28, 95% CI (−0.53, −0.03), P = .03	−1①	0	0	0	0	⊕⊕⊕○Moderate
Physical function
Duan et al 15	Mind-body therapy vs control	4 (219)	SMD = 0.46, 95% CI (0.19, 0.73), P = .0009	−1①	0	0	0	0	⊕⊕⊕○Moderate
Ma et al 32	Moxibustion vs control	7 (387)	MD = 10.46, 95% CI (4.41, 16.52), P = .0007	0	0	−1③	−1④	0	⊕⊕○○Low
Ni et al 22	Tai Chi vs rehabilitation training/psychological therapy: muscular strength of upper limbs	4 (330)	SMD = 0.49, 95% CI (0.16, 0.81), P = .003	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Tai Chi vs rehabilitation training/psychological therapy: upper limb function	5 (465)	SMD = 1.19, 95% CI (0.63, 1.75), P < .0001	−1①	−1②	0	−1④	0	⊕○○○Very low
Tao et al 40	Tai Chi vs control: vital capacity	3 (325)	MD = 339.53, 95% CI (78.5, 600.56), P = .01	−1①	0	−1③	−1④	0	⊕○○○Very low
Biological indices
Ni et al 22	Tai Chi vs psychological therapy/standard support: IL-6	2 (38)	SMD = 2.21, 95% CI (0.74, 3.68), P = .003	−1①	−1②	−1③	0	0	⊕○○○Very low
	Tai Chi vs health education/standard support: cortisol	2 (73)	SMD = −0.09, 95% CI (−0.16, −0.02), P = .01	0	−1②	0	0	0	⊕⊕⊕○Moderate
Adherence
Song et al 21	Yoga vs routine care	13 (1264)	OR = 0.55, 95% CI (0.4, 0.74), P < 0001	0	0	0	0	0	⊕⊕⊕⊕High
Adverse events
Liu et al 41	Kanglaite injection plus chemotherapy vs chemotherapy: leukopenia (grade Ⅱ-Ⅳ)	12 (1216)	RR = 0.29, 95% CI (0.22, 0,39)	−1①	0	0	−1④	0	⊕⊕○○Low
	Kanglaite injection plus chemotherapy vs chemotherapy: thrombocytopenia (grade Ⅱ-Ⅳ)	4 (674)	RR = 0.39, 95% CI (0.21, 0.71)	−1①	0	0	0	0	⊕⊕⊕○Moderate
	Kanglaite injection plus chemotherapy vs chemotherapy: phlebitis (grade Ⅱ-Ⅳ)	5 (291)	RR = 3.44, 95% CI (1.30, 9.15)	−1①	−1②	−1③	0	0	⊕○○○Very low
	Kanglaite injection plus chemotherapy vs chemotherapy: anemia (grade Ⅱ-Ⅳ)	8 (978)	RR = 0.54, 95% CI (0.42, 0.70)	0	0	0	−1④	0	⊕⊕⊕○Moderate

①Serious risk of bias; ②Suspicion of publication bias; ③The cumulative sample size was not enough, and/or 95% confidence interval was very wide; ④I²-value of the results was large ( ≥70%); ⑤I²-value of the results was very large ( ≥90%); KPS, Karnofsky Performance Status; QoL, Quality of Life; TCM-FEMT, Traditional Chinese Medicine Five-Element Music Therapy.

Effects on Cancer-Related Fatigue

Thirteen SRs^12-24 reported the effects of mind-body therapies (qigong, Tai Chi, yoga, dance, and music) on CRF (Table 5). Of these, significant reductions were observed with mind-body therapies versus control (three SRs, SMD = 0.59, 95% CI: 0.27, 0.92; moderate-quality; SMD = 1.03, 95% CI: 0.61, 1.44; moderate-quality; SMD = −0.47, 95% CI: −0.88, −0.06; low-quality)^12,13,15; yoga versus control (four SRs, SMD = −0.51, 95% CI: −0.79, −0.22; moderate-quality; SMD = −0.20, 95% CI: −0.34, −0.06; high-quality; SMD = 0.45, 95% CI: 0.09, 0.82; very low-quality; SMD = −0.31, 95% CI: −0.52, −0.10; low-quality)^18,19,23,24; yoga versus routine care (SMD = −0.52, 95% CI: –0.86, –0.18; moderate-quality) ²¹ ; qigong versus control (SMD = −0.93, 95% CI: −1.80, −0.06; very low-quality) ¹⁶ ; music versus control (SMD = −0.88, 95% CI: −1.49, −0.26; low-quality) ¹⁷ ; and a combination of Tai Chi with conventional supportive care versus conventional supportive care alone (SMD = −0.91, 95% CI: −1.30, −0.53; low-quality). ²⁰ Furthermore, one SR did not reveal any differences in CRF level between the Tai Chi and the control group. ²² One SR showed substantial reductions of CRF post-intervention (SMD = −0.89, 95% CI: −1.59, −0.19; very low-quality), ¹⁴ but no differences were observed at 3-month follow-up between the qigong and the control group. No SRs showed increased degrees of CRF following mind-body therapies.

Nine SRs^25-33 reported the effects of acupoint stimulation therapies (acupuncture, moxibustion, auricular point therapy, and acupressure) on fatigue in adults with cancer (Table 5). CRF was significantly improved with acupoint stimulation methods: acupressure or acupuncture versus standard treatment (SMD = −0.95, 95% CI: −1.66, −0.25; low-quality) or sham therapies (SMD = −0.28, 95% CI: −0.54, −0.01; moderate-quality) ²⁵ ; acupuncture versus control (4 SRs, SMD = −0.39, 95% CI: −0.55, −0.22; moderate-quality; SMD = −1.19, 95% CI: −2.25, −0.12; very low-quality; SMD = −1.26, 95% CI: −1.80, −0.71; low-quality; SMD = −1.06, 95% CI: −1.73, −0.40; low-quality)^27,28,30,33; acupressure versus control (SMD = −0.87, 95% CI: −1.19, −0.55; moderate-quality) ³¹ ; and moxibustion versus control (SMD = −1.17, 95% CI: −1.44, −0.90; moderate-quality). ³² Also, a combination of auricular point with standard care significantly reduced CRF compared with standard care alone (RR = 1.76, 95% CI: 1.42, 2.17; very low-quality). ²⁹ The addition of moxibustion to routine care provided a slight improvement in severe CRF (the score of Piper Fatigue Scale ≥7 points) compared with routine care alone (OR = 0.16, 95% CI: 0.07, 0.37; low-quality), but none to mild or moderate CRF. ²⁶

Two SRs^34,35 reported the effects of manipulative and body-based practices (massage and reflexology) on CRF (Table 5). Only one SR ³⁴ conducted a meta-analysis, showing that administration of massage versus usual care or sham massage improved CRF in adults with breast cancer (SMD = −0.61; 95% CI: −1.09, −0.13; low-quality). The other SR ³⁵ included 10 RCTs, 7 of which revealed that reflexology or massage significantly reduced CRF, and 3 reported no significant improvement following massage compared with routine treatment and care.

Three SRs^36-38 reported the effects of dietary supplements (ginseng, guarana, acetyl-l-carnitine, coenzyme Q10, etc.) on CRF (Table 5). One SR ³⁶ without meta-analysis included 9 RCTs, 5 of which showed that CRF could be reduced with supplements of guarana extract, branched-chain amino acids, coenzyme Q10, acetyl-l-carnitine, and chlorella extract, while the other 4 reported no differences between the dietary supplements and the control group. Two SRs^37,38 with meta-analyses revealed that ginseng provided significant improvements for CRF patients (SMD = 1.24; 95% CI: 0.05, 2.42; low-quality), but guarana did not (Table 5).

Three SRs^39-41 reported the effects of mixed CAM interventions or kanglaite injection (a kind of traditional Chinese medicine) on CRF (Table 5). One SR, ³⁹ involving 41 RCTs, did not conduct a meta-analysis due to clinical heterogeneity, of which 5 reported on CRF. Among these, 4 RCTs indicated that acupressure, massage, or reflexology therapy significantly reduced CRF. Another RCT with 286 participants showed no differences between the reflexology and the standard care or lay foot manipulation group. One SR ⁴⁰ reported the effects of acupuncture, Tai Chi, or qigong on CRF. Response rates of improving fatigue assessed by Piper’s Fatigue Scale doubled with acupuncture versus control (OR = 2.40; 95% CI: 1.40, 4.13; low-quality). CRF assessed by Brief Fatigue Inventory was slightly improved for acupuncture versus control (SMD = −1.41, 95% CI: −2.58, −0.24; very low-quality), as well as for mixed CAM therapies versus control (SMD = −1.19, 95% CI: −2.19, −0.18). One SR ⁴¹ showed that a combination of kanglaite injection and chemotherapy did not significantly reduce CRF compared with chemotherapy alone based on 2 RCTs.

Effects on Quality of Life

Seven SRs reported the effects of mind-body therapies (qigong, Tai Chi, yoga, and dance) on QoL (Table 6, Supplemental Appendix 5). Significant improvements were observed in QoL with qigong versus control at 12-week follow-up (MD = 6.57, 95% CI: 2.32, 10.84; very low-quality) ¹⁶ ; yoga versus control (SMD = 0.37, 95% CI: 0.11, 0.62; moderate-quality) ¹⁸ or non-active comparators (SMD = 0.27, 95% CI: 0.07, 0.46; moderate-quality) ¹⁹ ; Tai Chi versus usual supportive care at 3-month follow-up (SMD = 0.32, 95% CI: 0.07, 0.56; moderate-quality) ²⁰ ; and a combination of Tai Chi with usual supportive care either at 3-month follow-up or 6-month follow-up compared with conventional supportive care alone (SMD = 0.34, 95% CI: 0.26, 0.43; SMD = 0.40, 95% CI: 0.31, 0.49; both moderate-quality). ²⁰ One SR ²² also reported that Tai Chi positively affects both the physical (SMD = 0.34, 95% CI: 0.09, 0.59; moderate-quality) and psychological domains of QoL (SMD = 0.60, 95% CI: 0.12, 1.08; low-quality). Two SRs^15,23 showed no differences in the improvement of QoL respective of following qigong, yoga, or dance compared with control.

Six SRs reported the effects of acupoint stimulation therapies (acupuncture, auricular point therapy, and moxibustion) on QoL (Table 6, Supplemental Appendix 5). One SR, ²⁵ involving 8 RCTs, conducted a descriptive analysis of QoL, of which 6 indicated that acupoint stimulation significantly improved QoL. The addition of moxibustion to routine care has a positive effect on emotional function (SMD = 0.97, 95% CI: 0.72, 1.22; moderate-quality) and global QoL (SMD = 0.81, 95% CI: 0.20, 1.43; low-quality) compared with routine care alone, but no substantial improvements were observed in role function, cognitive function, and social function. ²⁶ QoL was also improved with acupuncture versus control (MD = 10.09, 95% CI: 7.26, 12.92; low-quality), ²⁷ auricular point therapy plus standard care versus standard care alone (MD = 7.34, 95% CI: 5.11, 9.57; low-quality), ²⁹ and moxibustion versus control (SMD = 0.88, 95% CI: 0.54, 1.22; high-quality). ³² A combination of auricular point therapy with standard care or moxibustion also significantly improved role function (MD = 8.99, 95% CI: 7.45, 10.53; MD = 9.39, 95% CI: 2.22, 16.56; both low-quality), emotional function (MD = 8.75, 95% CI: 6.51, 10.98; MD = 11.71, 95% CI: 4.41, 19.01; both low-quality), cognitive function (MD = 7.17, 95% CI: 4.85, 9.50; low-quality; MD = 7.96, 95% CI: 3.08, 12.85; moderate-quality), and social function (MD = 5.64, 95% CI: 4.17, 7.10; low-quality; MD = 11.01, 95% CI: 7.46, 14.56; moderate-quality) compared with standard care alone.^29,32 In addition, one SR ³³ revealed that acupuncture did not provide significant benefits in CRF compared with the control.

Three SRs reported the effects of manipulative and body-based practices (massage and reflexology) on QoL (Table 6, Supplemental Appendix 5). One SR ³⁴ with meta-analysis showed that massage did not improve QoL compared with usual care or sham massage in adults with breast cancer. Two SRs conducted descriptive analyses of QoL. Four studies of one SR ³⁵ reported QoL, of which one revealed that reflexology or massage substantially improved QoL at 6-week follow-up. In contrast, the other 3 reported no statistically significant differences between the massage or reflexology and the control group. Five studies of the other SR showed increased QoL in the massage or reflexology group, and 7 other studies reported no significant differences. ³⁹

Two SRs reported the effects of mixed CAM interventions or kanglaite injection on QoL (Table 6, Supplemental Appendix 5). Mixed CAM therapies versus control (SMD = 0.97, 95% CI: 0.18, 1.76; very low-quality) or massage versus control (SMD = 1.31, 95% CI: 0.21, 2.41; very low-quality) significantly improved QoL, which was assessed with cancer-specific QoL scales, while acupuncture or qigong did not. ⁴⁰ Acupuncture versus control provided significant improvements in QoL, assessed with means of the Karnofsky Performance Scale (KPS) (MD = 4.55, 95% CI: 0.75, 8.34; very low-quality) or remission rate of KPS (OR = 2.32, 95% CI: 1.26, 4.27; low-quality). ⁴⁰ The number of patients with a ≥ 10-point improvement in the KPS score was 2-fold for kanglaite injection plus chemotherapy versus chemotherapy alone (RR = 2.05, 95% CI: 1.60, 2.64; low-quality). ⁴¹

Effects on Depression

Thirteen SRs reported depression (Table 6, Supplemental Appendix 5). Of these, three^15,18,23 demonstrated significant reductions in depression following yoga compared with control, one ⁴⁰ showed that traditional Chinese medicine 5-element music therapy relieved depression compared with control. Eight SRs revealed that administration of qigong, Tai Chi, acupuncture, or massage did not improve depression. Five of 41 RCTs in the other SR ³⁹ also reported depression. Among these, 3 showed significant improvements following massage compared with visits by staff or usual care and attention, and 2 showed none.

Effects on Anxiety

Eight SRs reported anxiety (Table 6, Supplemental Appendix 5). Significant improvements were observed with yoga versus control^15,18; acupuncture versus control ²⁷ ; and mixed CAM therapies versus control. ⁴⁰ Three SRs^16,33,34 showed no differences in the alleviation of anxiety following qigong, acupuncture, or massage compared with control. Six studies in another SR ³⁹ also reported anxiety. Among these, 3 showed significant remissions of anxiety in the massage, reflexology, or acupressure group.

Effects on Sleep

Nine SRs reported sleep (Table 6, Supplemental Appendix 5). Sleep quality was significantly improved with CAM therapies: mind-body therapies or yoga versus control ¹⁵ ; Tai Chi versus control ²² ; and acupuncture versus control.^27,40 One review ¹⁴ revealed that qigong had a positive benefit in sleep post-intervention, but no differences were observed at 3-month follow-up. Three SRs^18,20,32 showed no improvements in sleep following yoga, Tai Chi, or moxibustion compared with control. Five studies in another SR ³⁹ without meta-analysis indicated that massage versus control or acupressure versus sham acupressure significantly improved sleep.

Effects on Pain

Seven SRs reported pain (Table 6, Supplemental Appendix 5). Significant reductions were observed in pain with acupuncture versus control,^27,40 and moxibustion versus control. ³² A combination of kanglaite injection and chemotherapy significantly improved chest pain compared with chemotherapy alone. ⁴¹ Two SRs^28,34 revealed no significant differences between the acupuncture or massage group and the control group. One SR ³⁹ did not conduct a meta-analysis, of these, 11 studies showed improvements in pain following massage, reflexology, or acupressure compared with control.

Effects on Body Mass Index

Four SRs reported body mass index (BMI) (Table 6, Supplemental Appendix 5). Mind-body therapies versus control ¹⁵ or Tai Chi versus control ⁴⁰ conspicuously reduced BMI. Two SRs^16,20 showed no statistical differences in the improvement of BMI following qigong or Tai Chi compared with control.

Effects on Physical Function

Six SRs reported physical function (Table 6, Supplemental Appendix 5). Mind-body therapies or moxibustion versus control significantly improved physical function.^15,32 Tai Chi provided considerable enhancements in upper limb function, muscular strength of upper limbs, and vital capacity.^22,40 In addition, 3 SRs^15,18,28 showed no significant differences in physical function following yoga or acupuncture compared with control.

Effects on Biological Indices

Three SRs reported biological indices (Table 6, Supplemental Appendix 5). Tai Chi significantly improved cortisol and IL-6 compared with control. ²² The other 2 SRs^16,34 showed no significant differences in cortisol following qigong, Tai Chi, or massage compared with control.

Adherence

Two SRs reported adherence to the yoga intervention (Table 6, Supplemental Appendix 5). One SR ²¹ with meta-analysis showed considerable differences in the adherence between the yoga and the routine care groups. In the other SR ²⁴ without meta-analysis, 4 studies showed high adherence (>80%), and 3 revealed low adherence (could not be described as a percentage).

Adverse Events

Seventeen SRs reported adverse events (Table 6, Supplemental Appendix 5). One SR ³¹ with meta-analysis indicated that no substantial differences were observed in the occurrence of adverse events between the acupressure and the control group. The other SR ⁴¹ also conducted a meta-analysis, showing that kanglaite injection might increase the incidence of grade II to IV phlebitis, and decrease the incidence of grade II to IV leukopenia, thrombocytopenia, and anemia. The mind-body therapies-related adverse events reported in the SRs were transient back spasms, panic attack and foot pain, transient muscle soreness, unilateral hip pain, flu-like symptom, leg cramps, swelling of the hand, stomach pain, shoulder pain, and side pain.^{13,15,16,18,24}

Adverse events induced by acupuncture or acupressure were pain, bruising, blood stasis, pruritus, or bleeding at the intervention site, headache, palpitations, fatigue, dizziness, nausea, and feeling of discomfort.^25-27,30,33 The auricular point-related adverse events reported in the SRs were pain, discomfort, or itchiness in the stimulated ear. ²⁹ The moxibustion-related adverse events were mild blisters, erythema, and slight itching. ³² The massage-related adverse events reported in the SRs were bruising, back pain, and increased blood pressure, which were resolved in a later massage session. ³⁵ Adverse events attributable to dietary supplements were palpitation, tachycardia, nausea, vomiting, insomnia, anxiety, diarrhea, xerostomia, dry skin, headache, neuropathy, leukopenia, anemia, and arthralgia.^36,38 No serious adverse events were reported in all included SRs.

Discussion

Main Findings

The primary results are presented in our published review. All the CAM interventions, except for kanglaite injection, showed significant potential to reduce CRF. Based on the included SRs, yoga was recommended to reduce CRF in adults with cancer, and the role of other CAM interventions requires further research. All the included CAM interventions showed significant improvements in QoL. As for psychological outcomes, yoga or acupuncture were effective in improving both depression and anxiety; and qigong or music ameliorated depression. With regard to physical outcomes, yoga, Tai Chi, or qigong enhanced physical function and improved sleep; acupuncture might improve sleep. Improvement in pain was also evident with acupuncture, moxibustion, or massage. In addition, significant improvements were observed in BMI, IL-6 and cortisol levels with Tai Chi.

Comparison to Current Recommendations

Yoga (a kind of practice combining stretches, meditation, and breathing to relax the mind and body) is recommended by the latest international guidelines such as guidelines set by the National Comprehensive Cancer Network (NCCN), ² American Society of Clinical Oncology (ASCO), ⁴² and European Society for Medical Oncology (ESMO), ⁴³ to treat CRF in patients either undergoing active cancer treatment or post-treatment. Our study provides high-quality evidence which suggests that yoga improves CRF and increases adherence. In another study evaluating the effects of non-pharmacological interventions on CRF, yoga showed the highest impact in cancer patients after treatment. ⁴⁴ Based on current evidence, the role of yoga as an effective treatment in adults with CRF is clear, but establishing the time and frequency of therapy requires further investigation.

Our review shows that qigong and music may reduce CRF, but the quality of evidence is low. Based on the current guidelines, the recommendation of qigong or music to reduce CRF still needs more research, especially in the post-treatment participants.

Tai Chi (a form of exercise that combines meditation, slow body movements, and controlled breathing) does not appear to be a viable option to treat CRF in the current guidelines. Our review provides moderate-quality and conflicting evidence for Tai Chi relieving CRF. Another study also summarized previous SRs, indicating that Tai Chi might have a beneficial effect on fatigue in patients with breast cancer. ⁴⁵ A more robust conclusion needs to be further assessed through well-designed and well-conducted clinical trials.

Acupuncture is widely used worldwide to promote health and improve disease-related symptoms. Our study provides moderate-quality evidence suggesting that acupuncture relieves CRF. Furthermore, one study ⁴⁶ conducted an overview to assess the effects of acupuncture-related therapies on CRF, showing potential advantages in terms of efficacy and safety. Another study ⁴⁷ focused on the impact of acupuncture on cancer-related conditions, which also indicated its benefits for cancer survivors with fatigue. Based on current evidence, acupuncture is recommended for the treatment of CRF in cancer survivors.

Moxibustion, acupressure, and auricular-point therapy are also used to improve health by stimulating specific points. Our study revealed moderate-quality evidence suggesting that moxibustion and acupressure reduce CRF, and very low-quality evidence suggesting that the auricular point therapy may relieve CRF. However, they are not mentioned in any guidelines, probably due to few studies.

Massage, a type of therapy that uses the hands to manipulate the soft tissues of the body, is recommended by the NCCN guidelines to reduce CRF in patients undergoing active cancer treatment, while the ASCO guidelines suggest that additional research is needed to verify the effects. Our study provides low-quality evidence suggesting that massage alleviates CRF. However, whether massage significantly improves CRF compared with routine treatment and care is uncertain. No other study has performed an overview of previous SRs to assess the effects of massage on CRF. Given the low methodological quality and small number of primary studies, no definite conclusions can be drawn. Reflexology applies pressure at specific points on the hands or feet. Some clinical trials have revealed that reflexology relieves CRF, ³⁵ while others observed no significant differences between reflexology and lay-foot manipulation or standard care, ³⁹ leading to confounded conclusions. Based on the included SRs, massage and reflexology showed great potential for reducing CRF, which deserves further research.

Dietary supplements, including various products such as vitamins and herbs, are common non-pharmacological strategies used for patients with cancer. Ginseng, a herb, has been confirmed to have immunomodulatory properties. ⁴⁸ Our study provides low-quality evidence suggesting that ginseng reduces CRF. However, current international guidelines have not reached a consensus on the use of ginseng because of its inconsistent effects. The use of acetyl-L-carnitine, coenzyme Q10, and guarana is not recommended to control CRF in the current guidelines due to the lack of related studies. A SR and meta-analysis revealed that guarana does not provide a significant reduction in CRF compared to placebo. ³⁸ Larger RCTs are needed to identify the effects of dietary supplements on CRF.

Our review provides moderate-quality evidence to suggest that yoga improves depression and anxiety, and very low-quality evidence suggests that music relieves depression. Music therapy is a kind of creative outlet. In traditional Chinese medicine, it is often used according to the Five Elements theory, which may improve the efficacy of general music therapy. Current guidelines recommend music to alleviate anxiety and depression, but extensive clinical studies are needed to evaluate the potential effects of yoga.

CRF and sleep disturbances were also highly correlated. ⁴⁹ Current guidelines recommend yoga or Tai Chi to improve sleep, which was also shown in our study. Furthermore, our review also provides high-quality evidence to suggest acupuncture to improve sleep. Another meta-analysis revealed that acupuncture significantly improved sleep in patients with breast cancer receiving active cancer treatment. ⁵⁰ Therefore, the effects of acupuncture on sleep require further investigation. Pain is a risk factor of CRF. ³ Our review provides moderate-quality evidence to suggest that acupuncture improves pain, and low-quality evidence suggests that moxibustion relieves pain. Current guidelines recommend physical modalities (massage, acupuncture, acupressure, etc.) for improving pain; however, the effects of moxibustion on pain need to be further evaluated in more clinical trials.

Cortisol dysregulation is related to cancer development and QoL, and abnormalities in cortisol rhythm may lead to CRF. ⁵¹ IL-6, a pivotal inflammatory cytokine, is associated with cancer development and cancer-related symptoms—such as fatigue, anxiety, and depression. ⁵² BMI is also considered to be an important prognostic indicator of cancer. Tai Chi was the only intervention regulating cortisol, IL-6, and BMI in our study; it may also improve cancer-related symptoms or cancer prognosis. However, no differences in IL-6 levels were observed in another meta-analysis between the Tai Chi and support therapy groups. ⁵³ Current guidelines do not refer to these findings, which should be confirmed by further studies.

Kanglaite injection is a traditional Chinese medicine, and its active substance is extracted from the herbal semen coicis. The addition of kanglaite injection to chemotherapy does not reduce CRF compared with chemotherapy alone, but improves QoL and some physical outcomes (pain, nausea, and vomiting), and decreases the incidence of adverse events. ⁴¹ However, no current international guidelines recommend kanglaite injection for the treatment of cancer or cancer-related symptoms because of the primary low-quality studies.^42,43

Strengths and Weaknesses of the Review

CAM is a broad set of medical products and practices which is not part of the dominant healthcare systems. The prevalence for “current use” and “past use” of CAM in patients with cancer is around 40% and 43%, respectively, reported by a SR and meta-analysis involving 18 countries. ⁵⁴ It also shows an increasing trend of CAM use in cancer from the 1970s to date. CAM is often used to lessen the side effects of cancer treatments and improve cancer-related symptoms, especially fatigue. The effects of CAM on CRF have been assessed in many studies,^12-14 but given the mixed results, it remains obscure to determine effective interventions based on evidence-based medicine.

The goal of this study was to summarize the evidence to date on CAM interventions in the treatment of CRF by conducting an overview without language limitations. Only SRs of RCTs were included to reduce the risk of bias. As far as we know, this is the first comprehensive review of the association between CAM interventions and fatigue outcomes of adults with cancer.

Thirty SRs were included in our review. Only one SR was assessed as “low” or above methodological quality with the AMSTAR-2 tool, and 19 SRs were deemed low bias risk from the ROBIS tool. As for the reporting integrity evaluated by the PRISMA checklist, it was regrettable that no SR reported all 27 items. The strength of evidence was rated by the GRADE tool. There were 5 pieces of high-quality evidence (CRF, QoL, sleep, and adherence) and 51 pieces of moderate-quality evidence, which included all the outcomes except adherence.

Notwithstanding the above assessments, there was considerable heterogeneity in CRF between the included SRs. The reasons were likely multifaceted: (1) The types of cancer were multiple. Nineteen of 30 SRs included patients with mixed types of cancer; (2) The timing of outcomes assessment varied and was not clarified in many SRs, such as post-intervention, 3-month follow-up, 6-month follow-up, etc.; (3) The approaches of outcomes assessment were diverse. For instance, there were 24 different scales to evaluate CRF in the included SRs. The heterogeneity may have confounded the overall efficacy of CAM interventions in the treatment of CRF, as well as other cancer-related symptoms. Therefore, future high-quality studies are required to verify the effects.

Clinical and Research Implications

Our review shows that yoga and acupuncture improve CRF in adults with cancer, which is also recommended by the current international guidelines. Some studies have revealed that acupuncture for at least 4 times or yoga for at least 6 weeks has substantial effects on CRF.^24,30 However, no consensus has been established on the unified criteria for the course or frequency of treatment. It should be well-designed in future RCTs in accordance with previous studies and be considered as a stratification factor when analyzing the results in future SRs.

Although massage relieves CRF based on one meta-analysis that included 5 RCTs ³⁴ and the use of qigong, Tai Chi, music, moxibustion, acupressure, auricular point therapy, or dietary supplements of ginseng in reducing CRF shows some favorable results, their recommendation has not yet been agreed upon in current guidelines because of the low quality of the methodology and few studies. Considering these shortcomings, conducting high-quality and extensive clinical research in the future is essential.

Besides the interventions mentioned above, other CAM therapies, such as reiki, imagery, and hypnosis, were not assessed in our study because related SRs were not found. For example, hypnosis, a type of mind-body therapy, can make people feel calmer and more focused by channeling them into a trance-like state. A SR of 13 RCTs showed that hypnosis improves CRF and pain. ⁵⁵ Therefore, the effects of other CAM interventions on cancer-related symptoms warrant further investigation.

Furthermore, our study clarifies that few specific indicators or biomarkers to assess CRF exist, which is worth pursuing in the future. For example, inflammation is supposedly a pivotal contributor to CRF, and one study evaluated the changes in IL-6 and IL-8 to explore the correlation between anti-inflammatory and CRF reduction. ⁵⁶ Moreover, many RCTs using CAM interventions, such as yoga and massage, cannot be designed with double-blinding, and CRF outcomes are mostly evaluated with subjective scales. Therefore, pursuing more specific indicators or biomarkers may reduce the risk of bias.

Conclusion

This overview demonstrates that CAM interventions, particularly yoga, Tai Chi, acupuncture, moxibustion, acupressure, massage, and dietary supplements of ginseng, improve CRF in adults with cancer. The roles of qigong, music, auricular point therapy, other dietary supplements, and other traditional Chinese medicines in CRF need further evaluation, as some data show encouraging results. The evidence for improvements in QoL, depression, anxiety, and sleep was more robust than that for other outcomes (pain, physical function, and BMI) and biological indices (cortisol and IL-6). However, given the heterogeneity of the included SRs and the low quality of original RCTs, conducting further research with rigorous methodological designs and sufficient sample sizes is still necessary. Additionally, performing SRs with high quality and reporting of comprehensive findings is essential.

Supplemental Material

sj-docx-1-ict-10.1177_15347354231188947 – Supplemental material for The Role of Complementary and Alternative Medicine on Cancer-Related Fatigue in Adults: An Overview of Systematic Reviews

Supplemental material, sj-docx-1-ict-10.1177_15347354231188947 for The Role of Complementary and Alternative Medicine on Cancer-Related Fatigue in Adults: An Overview of Systematic Reviews by Peijin Li, Qian Wang, Lixing Liu, Qinglin Zhang, Rui Zhou, Yue Wang, Tingting Liu and Li Feng in Integrative Cancer Therapies