Mindfulness, behavioural activation and breathing

These three are the methods people meet most often after exercise: a mindfulness course or app, a suggestion to "get active", and a breathing exercise. Each has evidence behind it, and for each the useful question is how far that evidence goes. This lesson reads all three with the same questions: what did the comparison group get, how sure is anyone, and who was left out?

Mindfulness: the effect, sorted by comparison

Focus and Deep Work lesson 6 asked whether meditation trains attention. The first review below has a line on that too: "We found low evidence of no effect or insufficient evidence of any effect of meditation programs on positive mood, attention, substance use, eating habits, sleep, and weight."¹ This lesson asks about mood and anxiety, where the evidence is stronger, and it sorts that evidence by two things: who was studied, and what they were compared with.

Patients, against something that looked like treatment

Start with the review that used the hardest comparison. Goyal and colleagues, in 2014, looked at trials in "diverse adult clinical populations", and this course read their abstract.¹ They "identified randomized clinical trials with active controls for placebo effects", meaning comparison groups given something that looked like a treatment. Across 47 trials with 3,515 participants: "Mindfulness meditation programs had moderate evidence of improved anxiety (effect size, 0.38 [95% CI, 0.12-0.64] at 8 weeks and 0.22 [0.02-0.43] at 3-6 months), depression (0.30 [0.00-0.59] at 8 weeks and 0.23 [0.05-0.42] at 3-6 months), and pain (0.33 [0.03- 0.62]) and low evidence of improved stress/distress and mental health-related quality of life."¹

Two more sentences from the same abstract. "We found no evidence that meditation programs were better than any active treatment (ie, drugs, exercise, and other behavioral therapies)." And their advice to clinicians: "meditation programs can result in small to moderate reductions of multiple negative dimensions of psychological stress."¹ Hold on to the first of those. It comes back in the next section.

Volunteers, against nothing

The other two reviews studied people who were not patients, and compared them mostly with a passive control: a waiting list, or nothing at all. One sign convention changes here. Goyal reported improvement as a positive number; these reviews report it as a negative one, because distress went down. Both mean the programme helped.

Galante and colleagues, in 2023, pooled individual participants' data from 13 trials of mindfulness-based programmes, or MBPs, in 2,371 adults. This course read the abstract. The review took in only trials "comparing in-person, expert-defined MBPs in non-clinical settings with passive control groups", and found that "In comparison with passive control groups, MBPs reduced average distress between one- and six-months post-intervention with a small to moderate effect size (standardised mean difference (SMD) -0.32; 95% confidence interval (CI) -0.41 to -0.24". And: "Confidence in the primary outcome result is high."²

An earlier review by the same lead author, in 2021, is the wider picture: 136 trials in non-clinical settings, and this course read the abstract's methods-and-findings and conclusions.² Against no intervention, MBPs helped: "Compared with no intervention, in most but not all scenarios MBPs improved average anxiety (8 trials; standardised mean difference (SMD) = -0.56; 95% confidence interval (CI) -0.80 to -0.33". Against a nonspecific active control, an activity meant to match the attention and expectation rather than to treat anything, they improved depression in most but not all scenarios, with no statistically significant evidence for anxiety or distress. Against specific active controls, other interventions meant to help: "Compared with specific active control conditions, there is no statistically significant evidence of MBPs' superiority."

Its conclusion, whole, because both halves matter: "Compared with taking no action, MBPs of the included studies promote mental health in nonclinical settings, but given the heterogeneity between studies, the findings do not support generalisation of MBP effects across every setting. MBPs may have specific effects on some common mental health symptoms. Other preventative interventions may be equally effective."² Heterogeneity means the trials' results varied more than chance would explain.

Where mindfulness has a specific result

One use has its own evidence. Mindfulness-based cognitive therapy, MBCT, was tested for preventing relapse in people with recurrent depression "in full or partial remission", meaning their symptoms had gone or mostly gone. Kuyken and colleagues pooled individual data from 1,258 patients, comparing MBCT with "usual care and other active treatments, including antidepressants", and this course read the abstract: "patients receiving MBCT had a reduced risk of depressive relapse within a 60-week follow-up period compared with those who did not receive MBCT (hazard ratio, 0.69; 95% CI, 0.58-0.82). Furthermore, comparisons with active treatments suggest a reduced risk of depressive relapse within a 60-week follow-up period (hazard ratio, 0.79; 95% CI, 0.64-0.97)."³

A hazard ratio compares rates over time. At 0.69, at any point in the 60 weeks, relapse ran at about 0.69 times the comparison group's rate, roughly 31% lower. The authors conclude that MBCT "appears efficacious as a treatment for relapse prevention for those with recurrent depression, particularly those with more pronounced residual symptoms", meaning people still carrying some symptoms.³

It isn't evidence for mindfulness as a general stress tool, and the authors do not claim it is.³

"Comparable to antidepressants"

In 2018, a group of mindfulness researchers wrote a critique of their own field, "Mind the Hype". This course read its abstract, introduction and three of its consensus sections. The abstract: "Misinformation and poor methodology associated with past studies of mindfulness may lead public consumers to be harmed, misled, and disappointed." And on one claim in particular: "MBIs are sometimes misleadingly described as "comparable" to antidepressant medications (ADMs)(Goyal et al., 2014)." MBIs are mindfulness-based interventions. They went on: "Given the absence of scientific rigor in much clinical mindfulness research (Davidson & Kaszniak, 2015; Goyal et al., 2014), evidence for use of MBIs in clinical contexts should be considered preliminary."⁴

The citation points at the review you've just read. This course read one paragraph of Goyal's Discussion, the one that makes the comparison: "These small effects are comparable with what would be expected from the use of an antidepressant in a primary care population, without the associated toxicities."¹ The next sentence sets them beside a separate analysis of antidepressants against placebo.

Why would the field's own researchers call that misleading? This course's reading of it has two steps.⁷ First, the comparison runs across studies. The meditation effects came from trials against active controls; the antidepressant figures came from other trials, of other people, against placebo. That's lesson 2's question, what did the comparison group get, asked of two sets of trials at once. Second, the abstract's own finding was "no evidence that meditation programs were better than any active treatment", and that's easy to hear as "as good as". It isn't the same claim. No evidence of a difference isn't evidence of no difference, because a small or noisy set of trials can miss a gap that is there. Showing "as good as" takes a trial that sets, in advance, how big a shortfall it will tolerate, and then shows the shortfall is smaller. That kind of trial is called a non-inferiority trial, and Goyal's point about side effects is worth keeping too, because the next trial measured them.

These critics are not opponents of mindfulness; they study it, which is what gives the critique its weight. A mindfulness teacher would add a fair point about scope. MBSR, mindfulness-based stress reduction, is a stress course, and its own standards exclude psychiatric disorder, as the next section shows. MBCT was built for relapse in depression, and NICE's group mindfulness row, quoted there too, describes a programme designed for people with depression.⁶

One trial since

Five years later, a trial tested the comparison for anxiety disorders. Hoge and colleagues randomised adults with diagnosed anxiety disorders to an eight-week MBSR course or escitalopram, a commonly prescribed antidepressant; this course read the abstract in full.⁵ The design was non-inferiority: "This randomized clinical trial (Treatments for Anxiety: Meditation and Escitalopram [TAME]) included a noninferiority design with a prespecified noninferiority margin." The main outcome was a clinician's rating of how severe each person's anxiety was, the "Clinical Global Impression of Severity scale (CGI-S)", and "The outcome assessments were performed by blinded clinical interviewer", so the rater did not know who had had which treatment.⁵

Of 276 recruited, 208 completed. "At end point, the mean (SD) CGI-S score was reduced by 1.35 (1.06) for MBSR and 1.43 (1.17) for escitalopram. The difference between groups was -0.07 (0.16; 95% CI, -0.38 to 0.23; P = .65), where the lower bound of the interval fell within the predefined noninferiority margin of -0.495, indicating noninferiority of MBSR compared with escitalopram."⁵

That sentence is dense, so take it in steps. The difference is MBSR's improvement minus the drug's, so a negative number means MBSR did a little less well: 1.35 against 1.43 is a gap of about 0.08, and the trial's own estimate is -0.07. The abstract doesn't label the 0.16 in brackets; the interval after it is the part to read. It runs from -0.38, MBSR somewhat worse, to 0.23, MBSR somewhat better. P = .65 says the two groups were not detectably different, which, as the last section showed, isn't the same as showing they're alike. The margin, -0.495, is the largest shortfall the trial agreed in advance to tolerate.

The result held in the other main way of counting: "Secondary intent-to-treat analyses using imputed data also showed the noninferiority of MBSR compared with escitalopram based on the improvement in CGI-S score."⁵ Intent-to-treat counts everyone randomised, not only those who finished, the analysis lesson 5 met in the exercise trials; "imputed" means missing scores were estimated.

On side effects the two differed a lot: "Of patients who started treatment, 10 (8%) dropped out of the escitalopram group and none from the MBSR group due to adverse events. At least 1 study-related adverse event occurred for 110 participants randomized to escitalopram (78.6%) and 21 participants randomized to MBSR (15.4%)."⁵ Those last percentages are of everyone randomised to each arm.

Here is the mindfulness evidence from this lesson in one place. The arrangement is this course's; the figures are the sources'.⁷

Who mindfulness isn't for

The same 2018 critique carries two sentences that matter more than any effect size for some readers: "The official standards of practice for MBSR exclude suicidality and the presence of any psychiatric disorder (Santorelli, 2014)." And: "The NIH states that "meditation could cause or worsen certain psychiatric problems" but does not provide any practice guidelines beyond a boilerplate disclaimer to "check with your doctor" before trying meditation (NCCIH, 2016)."⁴ NIH is the US National Institutes of Health, and NCCIH its National Center for Complementary and Integrative Health.

NICE's depression guideline places group mindfulness among its options for less severe depression, in a table where "all treatments in table 1 can be used as first-line treatments". Its row reads: "Uses a programme such as mindfulness-based cognitive therapy specifically designed for people with depression." And the same row adds a related caution, that it "May be difficult for people experiencing intense or highly distressing thoughts, or who find focusing on the body difficult."⁶

This course reads those sentences together this way.⁷ MBSR's own standards exclude suicidality and any psychiatric disorder, though it has since been trialled in anxiety disorders, as above. NICE's group programme is built for depression, with a caution for people whose thoughts are intense. And if you're having thoughts of suicide, a mindfulness course isn't the first step; a clinician is, and the callout at the top of this lesson has the numbers.

Behavioural activation

Behavioural activation, or BA, is the method of the three most people haven't heard of. Its Cochrane review holds 53 trials, all in adults with diagnosed or subthreshold depression, which is more clinical trials than either the mindfulness review of patients above (47) or the breathwork review below (12 on its main outcome); that count across the three is this course's.⁷

The Cochrane review defines it as "a brief psychotherapeutic approach that seeks to change the way a person interacts with their environment."⁶ NICE's depression guideline describes it as focusing on "identifying the link between an individual's activities and their mood", and says it "Does not directly target thoughts and feelings".⁶ So it works on what a person does rather than on what they think.

This course's gloss on why that might help is that doing can come before feeling like doing.⁷ Low mood tends to make people wait until they feel like it, and BA reverses the order: plan the activity, do it, and notice what happens to mood. That gloss is as far as this course can go. It has no BA manual or protocol among its sources, so it can't tell you what a session contains or how activities are chosen, and it will not guess. Habits and Self-Discipline lesson 1 defined a habit as an action triggered by a context rather than by a decision in the moment. BA shares the bet that you don't have to wait to feel like it, but it plans activities on purpose rather than making them automatic.

The review, by Uphoff and colleagues, included 53 studies with 5,495 participants. This course read its abstract.⁶

• Against treatment as usual: "moderate-certainty evidence that behavioural activation had greater short-term efficacy than treatment as usual (risk ratio (RR) 1.40, 95% confidence interval (CI) 1.10 to 1.78; 7 RCTs, 1533 participants), although this difference was no longer evident in sensitivity analyses using a worst-case or intention-to-treat scenario." Here the risk ratio counts a good outcome, so above 1 favours BA. A sensitivity analysis re-runs the numbers under different assumptions to see if the result survives. The review does not define "worst-case" in the abstract; this course takes it to mean counting people who dropped out as not improved.⁷ Intention-to-treat counts everyone who was randomised.
• Against CBT: "moderate-certainty evidence suggesting no evidence of a difference in short-term treatment efficacy between behavioural activation and CBT (RR 0.99, 95% CI 0.92 to 1.07; 5 RCTs, 601 participants)."
• Against other therapies and medication: BA "was more effective than humanistic therapy (RR 1.84, 95% CI 1.15 to 2.95; 2 RCTs, 46 participants; low certainty) and medication (RR 1.77, 95% CI 1.14 to 2.76; 1 RCT; 141 participants; moderate certainty), but both of these results were based on a small number of trials and participants." Humanistic and psychodynamic therapy are two other schools of talking therapy.
• Against a waiting list: "Compared with waiting list, behavioural activation may be more effective, but there were fewer data in this comparison and evidence was of low certainty (RR 2.14, 95% CI 0.90 to 5.09; 1 RCT, 26 participants)."
• Against placebo: "No evidence on treatment efficacy was available for behavioural activation versus placebo and behavioural activation versus no treatment."

And their conclusion, in its own terms: "behavioural activation may be more effective than humanistic therapy, medication, and treatment as usual, and that it may be no less effective than CBT, psychodynamic therapy, or being placed on a waiting list. However, our confidence in these findings is limited due to concerns about the certainty of the evidence."⁶

NICE places BA in more than one spot. Its guided self-help for depression can use "structured behavioural activation (BA)" materials, supported by a trained practitioner (lesson 3's table). Group BA has its own row in Table 1, for less severe depression, third of eleven, and individual BA is third in Table 2, for more severe depression. Table 1's heading says its order folds in cost and how easy a treatment is to provide, so the order is not a ranking by effect size.⁶

Breathing

A breathing exercise costs nothing and takes a minute, which is why so many apps offer one. The evidence base is small. Fincham and colleagues, in 2023, pooled randomised trials of breathwork "compared to non-breathwork controls", and this course read the abstract: "The primary outcome self-reported/subjective stress included 12 randomised-controlled trials (k = 12) with a total of 785 adult participants. Most studies were deemed as being at moderate risk of bias. The random-effects analysis yielded a significant small-to-medium mean effect size, g = - 0.35 [95% CI - 0.55, - 0.14]".⁸ A random-effects analysis allows the true effect to differ from trial to trial, and k is the number of trials. Effects on anxiety and depression were "similar".

The authors end on two sentences that belong together: "Overall, results showed that breathwork may be effective for improving stress and mental health. However, we urge caution and advocate for nuanced research approaches with low risk-of-bias study designs to avoid a miscalibration between hype and evidence."⁸

Two things to keep. The outcome is self-reported stress; nothing in this review's main result measures a change in the body. And this course has no source on any specific breathing technique, so it does not describe one as evidence-based. Lesson 4's applied relaxation, twelve to fifteen sessions with a trained practitioner, is a different thing from a breathing exercise.

Five things people get wrong

"Mindfulness works as well as antidepressants." In 2018 its own researchers called that a misleading summary. One trial since, for anxiety disorders, found an MBSR course not worse than a common antidepressant, with no placebo arm. A settled comparison would need more trials than one.

"Mindfulness is all hype." The review behind "comparable" found small to moderate effects against active controls in patients, and a review of volunteers rates its result with high confidence. MBCT lowered the risk of relapse in recurrent depression in remission, and NICE lists group mindfulness for less severe depression.

"Meditation is harmless." MBSR's own standards exclude people who are suicidal or have a psychiatric disorder, and NIH says meditation "could cause or worsen certain psychiatric problems". Even in Hoge's trial, 15.4% of those randomised to MBSR had at least one study-related adverse event, far fewer than on the drug but not none.

"Behavioural activation is just keeping busy." It's a structured therapy focused on the link between activities and mood, which did about as well as CBT in the trials that compared them.

"Breathing exercises lower your stress hormones." The main meta-analysis measured self-reported stress, in 785 people across 12 trials, most at moderate risk of bias.

Practice

Connections

Back. Lesson 2's question, what did the comparison group get, sorts the mindfulness evidence into its layers and explains the trouble with "comparable". Lesson 5 met intention-to-treat. Focus and Deep Work lesson 6 met meditation for attention; lesson 3 of this course placed guided self-help, which is where NICE puts BA's self-help form.

Forward. Lesson 7 is three popular ideas, resilience, a ratio and a mindset, where the distance between the evidence and the claim is widest of all.

Go deeper

• Van Dam and colleagues, "Mind the Hype", 2018, free at PubMed Central. This course read the abstract, introduction and three of the consensus sections. A critique of a field written by people who work in it.
• Galante and colleagues, 2023, free at PubMed Central. This course read the abstract. Its methods paragraph shows exactly which programmes and people the "high" confidence covers.
• Uphoff and colleagues, 2020, the Cochrane review of behavioural activation. Abstract only here. Its "Main results" paragraph shows how many comparisons rest on one or two trials.

Sources

1. M. Goyal and colleagues, "Meditation programs for psychological stress and well-being", JAMA Internal Medicine 174(3), 2014, doi 10.1001/jamainternmed.2013.13018, PMC4142584. Read: the abstract, and one paragraph of the Discussion (from PMC, 2026-09-24). Supports: the population, the active-control design, the effect sizes, the attention finding, no evidence of superiority to active treatments, the conclusion, and the "comparable" sentence.
2. J. Galante and colleagues, Nature Mental Health 1(7), 2023, doi 10.1038/s44220-023-00081-5, PMC7615230; and J. Galante and colleagues, PLoS Medicine 18(1), 2021, doi 10.1371/journal.pmed.1003481. Read: the 2023 abstract; the 2021 abstract's methods-and-findings and conclusions. Supports: the inclusion rule, the passive-control effect, its confidence and conditions, and the 2021 findings against no intervention and active controls and its whole conclusion.
3. W. Kuyken and colleagues, "Efficacy of Mindfulness-Based Cognitive Therapy in Prevention of Depressive Relapse", JAMA Psychiatry 73(6), 2016, doi 10.1001/jamapsychiatry.2016.0076, PMC6640038. Read: the abstract. Supports: the comparators, the relapse findings, the population and the residual-symptoms rider.
4. N. T. Van Dam and colleagues, "Mind the Hype", Perspectives on Psychological Science 13(1), 2018, doi 10.1177/1745691617709589, PMC5758421. Read: the abstract, introduction and the three "Consensus about..." sections. Supports: every quotation from it, including the MBSR exclusion and the NIH sentence.
5. E. A. Hoge and colleagues, "Mindfulness-Based Stress Reduction vs Escitalopram for the Treatment of Adults With Anxiety Disorders", JAMA Psychiatry 80(1), 2023, doi 10.1001/jamapsychiatry.2022.3679, PMC9647561. Read: the abstract in full. Supports: the design, outcome measure, sample, result, intent-to-treat confirmation, drop-outs and adverse events.
6. E. Uphoff and colleagues, "Behavioural activation therapy for depression in adults", Cochrane Database of Systematic Reviews 2020, CD013305, doi 10.1002/14651858.CD013305.pub2, PMC7390059 (read: the abstract); and National Institute for Health and Care Excellence, NG222, 2022, Table 1 with its heading and its guided self-help, group mindfulness and behavioural activation rows, recommendation 1.5.2, Table 2's row order, and recommendations 1.4.12 and 1.4.14 (read). Supports: BA's definitions, results and placement, the group mindfulness row and caution, and the advice on stopping an antidepressant.
7. This course's own constructions, labelled where they appear. The reading of the 2023 review's conditions, the two-step reading of "comparable", the setting of the 2018 critique beside the 2023 trial, the summary table, the reading of MBSR's exclusions for a reader, the count of trials across the three reviews, the gloss of BA and of "worst-case" are this course's.
8. G. W. Fincham and colleagues, "Effect of breathwork on stress and mental health", Scientific Reports 13(1), 2023, doi 10.1038/s41598-022-27247-y, PMC9828383. Read: the abstract. Supports: the comparator, the breathwork figures, the authors' overall reading and their caution.