Meditation-Related Adverse Effects

What it is

Published in Clinical Psychological Science in 2021, the study exists because of a gap in how the field collects harms. Trials of psychological interventions have historically asked about adverse effects in one line, at the end, in an open-ended way, if at all. Britton and colleagues built a structured 44-item instrument instead, covering specific categories of difficulty, and had it administered by an assessor independent of the teaching, to every one of the 96 participants who completed the programme. All of them had mild to severe depression, and all had completed one of three variants of an eight-week mindfulness-based cognitive therapy course.

Eighty-three per cent, 80 of 96, reported at least one side effect of any kind. Fifty-eight per cent, 56 of 96, reported at least one adverse effect of negative valence, meaning an experience they found unpleasant rather than merely unexpected. Of the 81 participants whose effects were rated for impact, 37 per cent, 30 people, reported at least one that had a negative impact on their functioning. For lasting effects the denominator is 78, and the rates depend entirely on what lasting means: 14.1 per cent reported bad effects persisting more than a day, 9.0 per cent more than a week, and 6.4 per cent more than a month. Those enduring cases were associated with signs of dysregulated arousal.

The result that generalises furthest is not any of those numbers. It is what happened when the same participants were asked the ordinary way. A single open-ended question of the sort trials normally use identified 26 of the 80 people who had a side effect. Asking politely whether anything had gone wrong missed roughly two-thirds of the cases the structured instrument found, in the same people, about the same eight weeks. Every published adverse-effect rate for every psychological intervention now has to be read as a joint product of what happened and how it was asked about.

Britton reads her own data conservatively, and the two readings available are worth keeping side by side. The alarming reading is that a majority of participants in a widely recommended wellbeing intervention experienced something unpleasant, that a third had their functioning affected, and that nobody had been counting. Britton's reading is that transient difficulty in meditation looks normative rather than dangerous, that participants evidently judged it tolerable since they completed the course, and that comparable rates, in the range of 3 to 14 per cent, turn up across psychotherapy generally. Both readings are defensible on these data. This site's position is that the second is closer to right about the numbers and that the first is closer to right about the practice, because the finding that survives is the methodological one: meditation is an active intervention with a side-effect profile, and until 2021 the field had no instrument capable of describing it. That is a statement about measurement, not a warning.

Britton discloses that she founded Cheetah House, a nonprofit that supports people who have had difficult experiences with meditation. That disclosure cuts in two directions at once and both should be stated. It gives her a prior about what the instrument would find. It is also the reason the instrument exists, since researchers without that exposure had not built one in forty years of trials.

In effect

Research: the same people, two instruments

Ninety-six participants, one eight-week programme, two ways of asking. The open-ended question found 26 cases. The 44-item interview, delivered by an assessor with no stake in the teaching, found 80. Nothing about the participants or their experiences changed between the two numbers. This is the cleanest available demonstration that a reported rate is partly a property of the questionnaire, and it applies well beyond meditation, to every trial, satisfaction survey and side-effect table built on an open prompt.

Everyday life: the participant nobody asked

Someone joins an eight-week course, finds that sitting with their attention on their breath brings up more agitation than calm in weeks three and four, assumes this is the practice working as intended because that is what the framing suggests, and is never asked a question specific enough to surface it. At the end, a facilitator asks whether anyone had any difficulties, and the room stays quiet. Nothing has gone wrong with anyone's honesty. The instrument was not built to catch it, and the social setting of a group course makes an open prompt about as low-yield as an open prompt can be.

Commerce: an industry with no reporting channel

Mindfulness reaches most people now through apps, workplace programmes and retreats rather than through clinical trials, and none of those routes has anything resembling adverse-event reporting. A medicine sold in the United Kingdom sits inside the MHRA's Yellow Card scheme, where patients and clinicians can report suspected reactions and the signal is aggregated. An eight-week course delivered by an employer has no equivalent, no denominator, and no obligation to look. That is not an accusation against any particular provider. It is a description of an asymmetry: the sector grew on the evidence base for benefits, which was collected, and not on an evidence base for harms, which until recently nobody had the means to collect.

What it does not say

It does not say meditation is dangerous, and this page should not be read as making that argument. There is no comparison group for the adverse-effect data reported here, so there is no way to know what fraction of these experiences would have occurred over the same eight weeks without the programme, in a sample of people who all had mild to severe depression to begin with.

It does not say meditation is harmful on balance. It reports costs. It says nothing about benefits, which are measured separately and in this population are the reason the programme was offered. A cost figure without a benefit figure is half a decision.

The authors' own reading is that these rates are unremarkable. Britton and colleagues place them alongside rates of roughly 3 to 14 per cent reported across psychotherapy generally, and note that participants continued and completed the course. Any use of these numbers that omits the authors' interpretation is quoting selectively.

It describes one kind of practice in one setting. Three variants of a supervised, structured, eight-week mindfulness-based cognitive therapy programme, taught by trained teachers to people with depression. It is not evidence about apps, about brief daily practice, about intensive retreats or about unsupervised practice, and the intensity differences between those are large.

The 96 are completers. People who left the programme are not in this denominator, which is a limitation in the direction of under-counting and is not resolved by these data.

Nothing here is medical advice, and nothing here is a reason to start or stop any practice or treatment. Anyone weighing that decision, particularly alongside a mental health condition, should be having the conversation with a clinician who knows their history rather than with a webpage.

---