Do burnout interventions work? What the evidence shows

The question

Burnout in medicine gets treated as a settled crisis with an unsettled cure. Institutions respond with a familiar menu: resilience workshops, mindfulness sessions, wellness committees, a subsidized yoga class. Clinicians respond to that menu with a well-documented contempt.

So the question is straightforward, and it has been studied: do these things work, and does it matter whether you aim them at the doctor or at the job?

What the meta-analysis found

The most-cited answer comes from a 2017 systematic review and meta-analysis in JAMA Internal Medicine, which pooled controlled trials of interventions to reduce physician burnout.

Two findings. Interventions produced small benefits, meaning something is better than nothing but nobody should expect a transformation. And, more usefully, the benefit was greater when the intervention was directed at the organization rather than at the individual physician.

That second point is the one worth carrying. It doesn't say individual interventions are useless. It says that if you have limited resources and you must choose, the evidence points at changing the work rather than at improving the worker's tolerance of it.

Why that ordering makes sense

It fits everything else we know. When clinicians are asked what drives their burnout, they name workload, administrative burden, documentation, and loss of control, which are all properties of the job. The Medscape mental health data puts bureaucratic tasks at the top of that list year after year.

If the cause is structural, an intervention aimed at the individual is treating a systems problem as a character problem. Worse, it carries an implicit accusation: if you had better coping skills, this wouldn't be happening to you. That's precisely why the wellness email lands the way it does.

The strongest recent result supports the same reading. The 2025 ambient AI scribe study found burnout falling from 51.9 percent to 38.8 percent in 30 days. It didn't teach anyone resilience. It took away the documentation.

The limits of this evidence, stated plainly

This is a research digest, so the caveats aren't decoration.

The pooled studies were heterogeneous in design, population, and intervention, which makes any single summary effect a rough instrument. Burnout is measured almost entirely by self-report on instruments that were built for research rather than for clinical decisions. Effect sizes were modest across the board, including for the organizational interventions that came out ahead. And there's a publication-bias problem in this literature, as in most.

A separate caution worth naming: a widely cited 2018 paper on burnout and patient safety, by an overlapping author group, was later retracted. That has no bearing on the 2017 interventions meta-analysis, which stands, but it's a reason to cite this literature carefully rather than reflexively.

What follows

If you're a clinician, the evidence gives you permission to stop treating your exhaustion as a personal deficiency. It's mostly the job.

If you're the one who can change the job, and that's usually a medical director, the evidence tells you where to spend. Reduce documentation. Fix the schedule. Restore some control over the work. Those are the levers with the better evidence behind them, and they're also the expensive ones, which is exactly why the yoga class keeps winning. See burnout in psychiatry and why the volume model burns clinicians out.