Hirshberg · Psychological medicine 2022 · pooled randomized controlled trials and prospective observational cohort · n=2429

Prevalence of harm in mindfulness-based stress reduction.

Cited 55 times in the scientific literature.

Level 2 - randomized trial

Combined analysis of pooled data from three randomized controlled trials and an observational cohort.

PubMed 32807249 · doi:10.1017/S0033291720002834 · record verified 2026-08-26

What was done

Researchers evaluated multiple indices of harm following Mindfulness-Based Stress Reduction (MBSR) using pre- and post-intervention data. The analysis combined an observational dataset from community health clinic MBSR classes (n = 2155) and participants from three randomized controlled trials (n = 156 MBSR; n = 118 waitlist control) conducted contemporaneously in the same city by the same clinic teachers. Primary outcomes were global psychological and physical symptoms; secondary outcomes included anxiety, depressive symptoms, discomfort in interpersonal relations, paranoid ideation, and psychoticism. Intent-to-treat analyses with multiple imputation measured overall symptom change, proportion of participants with increased symptoms, proportion with greater than a 35% symptom increase, and clinically significant harm for global psychological symptoms.

What was found

The abstract reports no numerical values, effect sizes, percentages, or test statistics. It found no evidence that MBSR led to higher rates of harm relative to waitlist controls on any primary or secondary outcome, and reported that community MBSR was significantly preventative of harm on many indices across multiple outcomes.

Why it matters

As mindfulness programs become increasingly common, empirical safety data are essential to evaluate concerns regarding potential psychological harm. This study indicates that standard MBSR does not elevate rates of symptom worsening compared to no treatment and may be protective against symptom deterioration.

Limits

The abstract provides no numerical data, confidence intervals, or specific p-values. The randomized comparison relied on a relatively small sample (n = 274 total) compared to the uncontrolled community cohort (n = 2155), used an inactive waitlist rather than an active control, and was restricted to pre- and post-test data without long-term follow-up.

Cited by