Published on 17 September 2026
Qigong, Tai Chi, Yoga: What Do Mind-Body Exercises Actually Do for Anxiety and Depression?
At a Glance
This meta-analysis pools 15 randomised trials and 1,351 adult participants, described by the authors as depressed or anxious but also including students and healthy older women, who practised qigong, tai chi or yoga for 8 to 24 weeks. Against passive control groups (waitlist, observation or no intervention), depressive symptoms fall further (Hedges’ g = −0.86, 95% CI −1.24 to −0.48), but heterogeneity across trials is massive (I² = 90%) and certainty is rated low. For anxiety, the effect is more modest, small to moderate by the authors’ own classification, and far more consistent (g = −0.38, 95% CI −0.53 to −0.23, I² = 19%), and the authors rate certainty as high. This contrast is what makes the study worth reading. Two caveats belong alongside it, though. Active-comparator trials were excluded by design: part of the effect may reflect the setting, the group and the expectation rather than the practice itself. And participants on psychotropic medication were excluded: these trials evaluate the practice alone, not its addition to a treatment.
Background
Many anxious or depressed patients ask what they can do alongside their treatment, and some are already practising yoga or tai chi. Physical exercise in general has been extensively meta-analysed, including previously on this site regarding exercise in bipolar disorder. Mind-body practices have a specific feature: they combine slow movement, breathing and attention, and their own effect is hard to separate from that of the setting in which they are practised.
The question asked here is therefore a precise one. In adults with depressive or anxious symptoms, do mind-body exercises reduce those symptoms compared with no active intervention, and under what conditions?
The Study at a Glance
| Question (PICO) | |
|---|---|
| Population | |
| Adults presented by the authors as diagnosed with depression or an anxiety disorder (ICD or DSM), although the trial table also includes students, healthy older women and pregnant women, with baseline levels classed as normal in 4 of 15 depression comparisons and 2 of 11 anxiety comparisons, and the authors themselves acknowledge that most trials did not specify diagnostic status. Participants on psychotropic medication excluded. Mean ages from 20.29 to 66.64 years per the trial table (21.1 to 66.64 years per the text). 1,351 participants, 787 in the exercise groups and 564 in the control groups | |
| Intervention | |
| Qigong, tai chi or yoga practised alone, sessions of 20 to 90 minutes, from once a week to daily, over 8 to 24 weeks | |
| Comparator | |
| Passive controls only: waitlist, observation or no intervention, with no exercise or other therapeutic intervention | |
| Outcomes | |
| Depressive symptoms (CES-D, SDS, HAMD, DASS, HADS, SCL-90, POMS) and anxiety symptoms (SAS, STAI, DASS, HADS, SCL-90), per the scales listed in the trial table, expressed as Hedges’ g, random-effects model | |
| Design | |
| Systematic review and meta-analysis of 15 randomised trials published between 2010 and 2023 per the trial table (2010 to 2024 per the text), protocol registered on PROSPERO (CRD42024613769), search across 7 databases up to July 2024 (the 1st or the 6th depending on the section), trial quality assessed with the PEDro scale, certainty with GRADE |
Quality Check
| Criterion | Status |
|---|---|
| Registered protocol | Yes |
| FindingPROSPERO CRD42024613769, search across 7 databases | |
| Trial quality | Average |
| FindingPEDro score from 6 to 9 (mean 6.33); blinding of participants, therapists and outcome assessors in 1 of 15 trials, concealed allocation in 2; mostly self-reported measures | |
| Comparator | Passive only |
| FindingActive-comparator trials were excluded: none compares against a group activity, another form of exercise or an attention control | |
| Publication bias | Tested |
| FindingEgger’s test non-significant and trim-and-fill analysis stable, which is not enough to rule out bias with so few trials | |
| External validity | Caveat |
| Finding12 of 15 trials conducted in China, including 2 in Hong Kong (2 in the United States, 1 in India); disparate populations (students, older women, pregnant women, women in a drug rehabilitation centre, post-stroke patients); participants on psychotropic medication excluded | |
| Funding and competing interests | None declared |
| FindingThe authors declare they received no funding and have no competing interests; open-access article | |
The Results
| Outcome | Result |
|---|---|
| Depressive symptoms (14 trials) | g = −0.86 (95% CI −1.24 to −0.48), I² = 90%, 15 comparisons, 1,311 participants, GRADE certainty low |
| PEB readingLarge effect, unreliable the heterogeneity is explained neither by the subgroups nor by the meta-regression (R² = 5.9%) | |
| Anxiety symptoms (10 trials) | g = −0.38 (95% CI −0.53 to −0.23), I² = 19%, 11 comparisons, 1,012 participants, GRADE certainty high according to the authors |
| PEB readingSmall-to-moderate, consistent effect against a passive control, which caps our confidence | |
| Qigong subgroup | Depression g = −1.228 (95% CI −1.805 to −0.650; 8 comparisons, I² = 93%); anxiety g = −0.439 (95% CI −0.634 to −0.245; 5 comparisons) |
| PEB readingExploratory an indirect comparison between practices, not a ranking; for anxiety, exercise type is not a significant moderator (p = 0.365) | |
| Dose subgroups, depression | 3 sessions per week: g = −2.045 (95% CI −3.070 to −1.020; 2 comparisons); sessions of 31 to 60 minutes: g = −1.315 (95% CI −1.874 to −0.755; 8 comparisons) |
| PEB readingHypothesis, not a dosage a g above 2 based on two comparisons mainly signals fragility, and the meta-regression finds no effect of session duration, frequency or period | |
The finding on anxiety is the most useful, and the best supported. It is worth being clear, though, about what “high certainty” means here: there is probably a difference compared with doing nothing. What the meta-analysis does not tell us is whether these practices do better than a group walk, a relaxation class or any other regular, supervised activity. And the frequency or duration subgroups are not a dosage: they compare different trials against each other, with different populations and different practices.
Critical Appraisal
| Domain | Judgement |
|---|---|
| Search and selection | Correct, inconsistent reporting |
| FindingRegistered protocol, 7 databases, PRISMA method, but a variable search end date (1 or 6 July 2024), 13,891 or 13,901 records identified depending on the text or the flow diagram, participants described as diagnosed although several populations are not, and the control group of the two-arm Hua and Sun (2021) trial counted twice in the analyses | |
| Non-specific effects | Not controlled for |
| FindingExpectation, social contact, weekly structure: everything that accompanies the practice is folded into the effect | |
| Symptom measurement | Caveat |
| FindingMostly self-reported scales (the authors write “all”, but 3 trials use the HAMD), completed by participants who know what they are practising | |
| Heterogeneity, depression | Very high |
| FindingI² = 90%, not resolved by the subgroups (I² as high as 95% within some of them) nor by the meta-regression (R² = 5.9%, residual I² = 88%) | |
| Certainty for anxiety | High, debatable |
| FindingNo downgrade applied despite the absence of blinding in 14 of 15 trials and mostly self-reported measures, limitations the authors themselves note | |
| Follow-up | Not analysed |
| FindingMost trials had no follow-up after the programmes ended, and the meta-analysis does not examine whether the effect is maintained | |
| Fit of the conclusion | Overinterpreted |
| FindingThe abstract speaks of robust evidence for depression despite low certainty; the conclusion attributes the effects to practice parameters even though the meta-regression is negative, and for anxiety recommends qigong for at least 16 weeks in the most anxious patients, whereas the abstract points to tai chi, 8 weeks or less and normal baseline levels, with no significant moderator at all | |
Level of Evidence
PEB assessment: reasonable confidence for anxiety, low for depression, low for the practice parameters. What is demonstrated, with the caveats attached to a passive comparator, is a small-to-moderate reduction in anxiety symptoms relative to no intervention. What is suggested is a reduction in depressive symptoms, whose true magnitude is unknown. What remains a hypothesis is the superiority of qigong over other practices and the existence of an optimal frequency or duration, which the meta-regression does not confirm. On this point, we are more cautious than the authors.
The Colleague Test
What an experienced colleague would say if shown this study for two minutes, between two consultations.
“For anxiety, I’m happy to suggest it, the effect holds up from one trial to the next. For depression, as an add-on, and without promising anything. But the controls are waitlist or nothing: I don’t know whether it’s the practice itself or just going to a class several times a week. And I’m not going to prescribe three sessions of 31 to 60 minutes like a dosage.”
In practice: a regular mind-body practice is a reasonable add-on to suggest, especially for anxiety, provided it is presented for what it is and never substituted for an established treatment. The included trials enrolled participants who were not on psychotropic medication: the value of adding this to a pharmacological treatment was not tested here.
What You Can Do With This
- Suggest a regular mind-body practice to an anxious patient as an adjunct to their care, choosing whichever one they are most likely to stick with, keeping in mind that the trials evaluated the practice alone, in participants who were not on psychotropic medication.
- In depression, present it as a possible adjunct, never as an alternative to treatment, and monitor progress as usual.
- Answering a patient who asks whether yoga “treats anxiety”: for mind-body practices taken together, there is a measurable, small-to-moderate effect relative to doing nothing; for yoga alone, the 3 available comparisons show no significant effect (g = −0.205, 95% CI −0.495 to 0.084), which does not prove there is none; and whether these practices do better than another supervised activity is unknown.
- Be wary of programmes marketed with a precise dosage: the frequency and duration data in this meta-analysis are exploratory.
- The course of action is set out in the NICE decision tree for generalised anxiety and panic disorder.
Frequently Asked Questions
Should qigong be preferred over yoga or tai chi?
Not on the basis of this study. Qigong’s advantage in depression comes from a comparison between subgroups of different trials, not from trials that pit the practices directly against one another. For anxiety, exercise type is not a significant moderator.
What does “high certainty” mean for anxiety?
That the authors consider it very unlikely that new trials against passive controls would change the result much. It says nothing about how these practices compare with another group activity.
Do these results generalise beyond the countries where the trials were run?
With caution. Twelve of the fifteen trials were conducted in China, including two in Hong Kong, in settings where these practices are culturally more familiar.
Does the effect last after the practice stops?
We don’t know. According to the authors, most of the included trials had no follow-up after the programmes ended, and the meta-analysis does not examine whether the effect is maintained.
Annotated Bibliography
Source study. Ye Z, Xu Z, Wang X. A systematic review and meta-analysis of the effects of mind-body exercise on depressed and anxious individuals. PeerJ. 2026; 14: e20570. DOI 10.7717/peerj.20570 · PMID 41522512, identifier established in the PubMed registry on 10 September 2026. Published online on 6 January 2026. Funding: the authors declare they received no funding for this work. Competing interests: the authors declare they have no competing interests. Registration: PROSPERO CRD42024613769. Supplementary material consulted: raw data by comparison (sample sizes, means and standard deviations of the changes, for depression and anxiety) and the PRISMA 2020 checklist.