What the Evidence Actually Shows
Real numbers, honest comparisons, and the gaps
Two questions, often blurred into one
Ask whether mindfulness works and you get a shouting match. Ask the two real questions separately and the picture settles down quickly. Better than nothing? Yes, fairly reliably. Better than another decent treatment? Usually not, and that is a different and more useful thing to know.
The careful review everybody should know
Goyal and colleagues, writing in the journal JAMA Internal Medicine in 2014, screened over eighteen thousand citations and kept forty-seven randomised trials covering 3,515 people. Measured against active control conditions, meaning something else that also took the participant's time and attention, meditation programs produced effect sizes of 0.38 for anxiety and 0.30 for depression at eight weeks, and 0.33 for pain. Those are small to moderate effects, and that is the phrase the authors used.
The same review found low or insufficient evidence of any effect on positive mood, attention, sleep or weight, and no evidence that meditation programs beat any active treatment, including medicines, exercise and other behavioural therapies. Both halves are the finding. A course that quotes the first half and drops the second is selling something.
The bigger picture, ten years on
Goldberg and colleagues reviewed forty-four meta-analyses in 2022, covering 336 randomised trials and 30,483 participants. Mindfulness-based programs beat passive controls such as waiting lists across most comparisons. Against active controls the effects were, in their words, typically smaller and less often statistically significant. Against established evidence-based treatments they came out similar or better. The strongest and most consistent evidence was for depression, pain, smoking and addiction.
Beats a waiting list is a weak claim, because almost everything beats a waiting list. Matches a treatment we already trust is a strong claim. Whenever you read a mindfulness result, find out which one it is before you decide what it means.
The head-to-head trials worth naming
Hoge and colleagues, JAMA Psychiatry, 2023. Adults with anxiety disorders were randomised to MBSR or to escitalopram, a first-line medicine. In the comparison sample of 208 people, severity fell by 1.35 points with the course and 1.43 with the medicine, a difference of 0.07 that sat inside the prespecified margin. The side effect picture was not close: 78.6 percent of the medicine group reported study-related adverse events and 8 percent stopped because of them, against 15.4 percent in the course group with nobody stopping.
Cherkin and colleagues, JAMA, 2016. Three hundred and forty-two adults with chronic low back pain were randomised to MBSR, to cognitive behavioural therapy, or to usual care. At twenty-six weeks, meaningful improvement in function reached 60.5 percent with MBSR, 57.7 percent with the therapy and 44.1 percent with usual care. Both active treatments beat usual care, and they did not differ from each other.
Polusny and colleagues, JAMA, 2015. A hundred and sixteen veterans with posttraumatic stress disorder (PTSD) were randomised to MBSR or to an active group therapy. Symptom scores favoured MBSR by about six points at follow-up, and 48.9 percent showed clinically significant improvement against 28.1 percent. But the proportion who no longer met criteria for the diagnosis did not differ, 53.3 percent against 47.3 percent, and the authors themselves called the average improvement modest.
The cousin with the best relapse record
Mindfulness-Based Cognitive Therapy (MBCT) borrows the eight-week shape and aims it at one target: stopping depression coming back. Kuyken and colleagues, JAMA Psychiatry, 2016, pooled individual patient data from nine trials and 1,329 people. Relapse within sixty weeks ran at 38 percent with MBCT against 49 percent with usual care, a hazard ratio of 0.69, and 0.79 against other active treatments including maintenance antidepressants. Benefit was larger for people carrying more residual symptoms.
The family resemblance
- MBSR: eight-week group course, general stress and chronic illness, no diagnosis required.
- MBCT: same shape, built to prevent depressive relapse, with cognitive therapy elements added.
- Acceptance and Commitment Therapy (ACT): individual or group therapy, mindfulness in service of psychological flexibility and valued action.
- Dialectical Behaviour Therapy (DBT): mindfulness as one taught skills module inside a wider treatment for emotion dysregulation.
- Relaxation training: aims at lowering arousal directly, which is a different goal, and often the comparison condition in these trials.
Van Dam and fourteen co-authors published a critical review in 2018 arguing that fuzzy definitions, weak methods and overstatement in this field can leave the public harmed, misled and disappointed. Keep it beside you. Being enthusiastic about a practice is not a reason to be loose about its evidence.