Breathing exercises reduced stress across 12 trials, by a small-to-medium amount

Deliberate breathing exercises do reduce self-reported stress, and the size of the effect has been measured: pooled across 12 randomised controlled trials with 785 adult participants, breathwork produced a small-to-medium effect of g = −0.35 (95% CI −0.55 to −0.14; z = 3.32, p = 0.0009) compared with non-breathwork control conditions [s1]. Most of those trials were judged to be at moderate risk of bias [s1].

Both halves of that sentence belong in any honest summary, and the researchers who produced it made the same point more bluntly than most authors would.

What was pooled

The meta-analysis searched PsycInfo, PubMed, ProQuest, Scopus, Web of Science, ClinicalTrials.gov and ISRCTN up to February 2022, initially identifying 1,325 results [s1]. The primary outcome was self-reported or subjective stress, for which 12 randomised controlled trials with 785 adult participants qualified [s1].

The random-effects analysis returned a significant small-to-medium mean effect size of g = −0.35 (95% CI −0.55 to −0.14), indicating lower stress in the breathwork conditions [s1]. Heterogeneity was intermediate and approaching significance (χ²₁₁ = 19, p = 0.06, I² = 42%) — meaning the constituent trials were reasonably but not perfectly consistent with one another [s1].

Two secondary outcomes were analysed with larger sets of trials: self-reported anxiety across 20 studies (g = −0.32, p < 0.0001) and depressive symptoms across 18 (g = −0.40, p < 0.0001) [s1]. Heterogeneity was moderate and statistically significant for both, which is a weaker footing than the primary analysis [s1].

The authors' own warning

Papers rarely editorialise about their field. This one does, and the sentence is worth quoting because it is unusual: having reported significant effects across all three outcomes, the authors write that they urge caution and advocate for nuanced research approaches with low risk-of-bias study designs "to avoid a miscalibration between hype and evidence" [s1].

That is the authors of a positive meta-analysis telling readers not to over-read it. Most of the included studies were assessed as being at moderate risk of bias [s1], and breathwork trials carry a structural problem that no amount of pooling fixes: participants know whether they have been asked to breathe in a particular pattern, so expectancy cannot be blinded away.

What the clinical literature adds

A separate scoping review looked only at studies of breathwork in adults with an anxiety disorder diagnosed under DSM-5, and only where breathwork was used without other interventions bundled in [s2]. From 1,081 articles identified across PubMed, Embase and Scopus, 16 met those criteria [s2].

Across that set, a range of breathwork interventions yielded significant improvements in anxiety symptoms in patients with a clinical diagnosis [s2]. The review notes that anxiety disorders are the most common group of mental disorders but are often underrecognised and undertreated in primary care, and that dysfunctional breathing is a hallmark of anxiety disorders even though the mainstays of treatment do not address breathing [s2].

The inclusion rule that produced those 16 studies is worth noting, because it is stricter than usual: only clinical studies using breathwork without the combination of other interventions, in adult patients carrying a DSM-5 anxiety disorder diagnosis, were eligible [s2]. Most published work on breathing bundles it with meditation, yoga or psychotherapy, which makes it impossible to say which component produced any effect. Applying that filter to 1,081 candidate articles left 16 [s2] — a measure of how little of the field isolates breathing as the active ingredient.

The review also reports a finding that runs against tidy summaries: results around the role of hyperventilation in the treatment of anxiety were contradictory in several of the examined studies [s2]. Breathwork is not one intervention. Slow-paced breathing and deliberately fast, high-ventilation techniques are physiologically opposite manoeuvres, and the evidence does not treat them as interchangeable.

The limits, stated plainly

Sixteen clinical studies and 12 trials on the primary stress outcome is a small evidence base for a practice this widely promoted [s1] [s2]. The scoping review is a scoping review — it maps the nature and extent of the literature rather than pooling it statistically [s2]. Outcomes across both bodies of work are largely self-reported, over short follow-ups, in trials whose control conditions vary considerably.

Nothing here establishes how long any effect lasts, which specific technique produces it, or whether breathwork compares favourably with other stress interventions — none of the analyses reported a head-to-head comparison against an established treatment.

The calibrated version

Breathing exercises are not inert. Across the randomised evidence that exists, they were associated with lower self-reported stress, anxiety and depressive symptoms, at effect sizes in the small-to- medium range [s1]. In people with diagnosed anxiety disorders, the small clinical literature points the same way [s2].

What the evidence does not support is the stronger claim that a particular breathing protocol reliably produces a large change in stress. The people who did the pooling said as much themselves, and a reader is better served by their caution than by a rounder number. This article describes research findings and is not medical advice.