Review Comparative efficacy of various exercise therapies for chronic fatigue syndrome: A systematic review and network meta-analysis, 2025, Liao et al.

[SNT Gatchaman]

Comparative efficacy of various exercise therapies for chronic fatigue syndrome: A systematic review and network meta-analysis

Spoiler: Authors

Zhongxin Liao; Suhong Zhao; Sitong Fang; Jun Ren; Shoujian Wang; Lingjun Kong; Min Fang

To systematically compare the effectiveness of different exercise therapies for chronic fatigue syndrome (CFS), we conducted a systematic review and network meta-analysis of randomized controlled trials based on searches of nine databases up to February 19, 2025.

The review included 25 randomized controlled trials, with 20 trials (n = 2,831) eligible for network meta-analysis. Graded exercise therapy (GET) showed relatively superior short-term effects on fatigue (mean difference [MD]: −6.93, 95% confidence interval [CI]: −10.85 to −3.01; moderate certainty), depression (MD: −5.27, 95% CI: −7.38 to −3.16; low certainty), and anxiety (MD: −2.88, 95% CI: −5.10 to −0.66, low certainty) compared with waitlist at the end of treatments, with partial maintenance of effects at follow-up.

Other modalities, including Qigong, Yoga, strength/resistance training, and running showed modest benefits but failed to surpass the minimally important difference with confidence, and were supported by low/very low certainty evidence.

These findings support the short-term utility of GET in managing CFS symptoms. However, its broader clinical endorsement remains controversial, highlighting the need for further high-quality trials.

HIGHLIGHTS
• Graded exercise therapy showed moderate-certainty short-term benefit in fatigue

• Other exercise modalities showed possible benefit with low/very low evidence certainty

• Long-term effects remain uncertain

• Findings highlight the need for individualized exercise and rigorous future trials

Web | DOI | PDF | iScience | Open Access

 

[SNT Gatchaman]

Unfortunate to see this published in iScience. The PACE trial is assessed as low risk of bias. Larun et al. (2024 !!) is of course featured.

Adaptive pacing therapy, a patient-centered approach, may help individuals manage symptoms, yet its effectiveness remains uncertain due to a lack of robust supporting evidence. By contrast, exercise therapy (ET) directly targets physical deconditioning and is one of the most widely recommended treatments for CFS. The Cochrane systematic review indicated that ET may offer broader benefits across fatigue, physical function, and self perceived general health.

 

[SNT Gatchaman]

Despite these strengths, the clinical use of GET remains highly debated. Although our analysis indicates significant benefits of GET both immediately post intervention (MD: − 6.93) and at long-term follow-up (MD: − 7.06), its broader endorsement is controversial.

The UK National Institute for Health and Care Excellence 2021 guideline underscores that no curative treatment for CFS exists and instead promotes patient-led symptom management. It explicitly advises against recommending structured exercise programs with fixed incremental progression (e.g., GET), particularly those based on deconditioning or activity-avoidance models.

This position has been challenged by Professor Flottorp, who criticizes the guideline for its lack of robust evidence and methodological shortcomings, including issues in diagnostic definitions, study selection, data analysis, and evidence grading. […] A more fundamental source of controversy relates to safety concerns. Professor Vink, for instance, has highlighted critical flaws in the supporting evidence base, such as inadequately controlled study designs, reliance on subjective fatigue measures in unblinded trials, neglect of objective outcomes like the 6-min walk test, and insufficient reporting of adverse events, with some patients experiencing symptom exacerbation.

These divergent views underscore the need for rigorous, transparent evaluations of GET that integrate both subjective and objective outcomes, systematically assess harms as well as benefits, and emphasize the importance of professional oversight by trained physiotherapists to ensure individualized and safe implementation. One possible explanation for the relatively superior short-term effects of GET is its structured and progressive nature, which may help reduce physical deconditioning and fear-avoidance behaviors, thereby improving patients’ confidence and adherence to activity.

Everyone's a professor it seems. (I didn't see a statement on the use of generative AI in the paper.)

We would like to thank the Shanghai Municipal Central-Local Joint S&T Development Fund Program (YDZX20243100002004), the Shanghai Key Laboratory of Tuina Techniques on Musculoskeletal Disorders (24dz2260200), Three Year Action Plan for Shanghai to Further Accelerate the Inheritance, Innovation and Development of Traditional Chinese Medicine (ZY(2025–2027)-3-1-1), a special project of Shanghai Pudong new area health commission (PW2023E01), and Shuguang Hospital Open Competition Program (SGYYJBGS-001).

 

[TiredMathematician]

Not a single mention of PEM in the entire article. Useless.

 

[Utsikt]

The authors have demonstrated a complete lack of understanding of even the most basic research methodology.

They used Cochrane’s Risk of Bias tool. The tool has many flaws, but if let’s ignore those. Domain 4 relates to the measurements used, and includes e.g. these considerations:

Whether:

• the method of measuring the outcome was inappropriate;
• measurement or ascertainment of the outcome could have differed between intervention groups;
• outcome assessors were aware of the intervention received by study participants;
• (if applicable) assessment of the outcome was likely to have been influenced by knowledge of intervention received.

All of the studies were unblinded and all of them used subjective outcomes. Therefore, all of them should have been rated as a high risk of bias in D4 (more details here), and consequently all of them should be rated as a high risk of bias overall.

These are the ratings the authors gave:

 

[Turtle]

I really hope Professor Vink can enjoy his promotion, all the way from Shanghai, whether human error or AI.
I like the sound of it. Congratulations professor Marc Vink. You would have earned it with your work on warning against GET.
If the professor is not lasting, your contributions to keep others savely away from GET, will last and even in Shanghai people are warned.