Teach-back: the evidence behind a thirty-second check

AA 58-year-old woman with type 2 diabetes nods through the last two minutes of her appointment. The clinician has explained a dose change, a new blood pressure pill, and a glucose target. She smiles. She thanks him. She walks out. Three weeks later she returns hypoglycemic because she doubled the insulin instead of the metformin. None of this is unusual. It is, in fact, the scenario that researchers have been trying to interrupt for two decades with a single closing question: would you mind telling me, in your own words, what you are going to do when you get home? That question is the heart of the teach-back method, and the evidence behind teach-back communication is unusually consistent for a clinical habit that costs roughly thirty seconds.

For readers searching for what is teach-back, the technique is straightforward. After explaining a diagnosis, a medication change, or a discharge plan, the clinician asks the patient to restate the plan in their own words. The clinician then clarifies anything that was missed and asks again. The frame matters: the question is not a quiz of the patient but a check on the clarity of the explanation. Done well, it sounds like, *I want to make sure I explained this clearly, can you tell me what you will do when you get home?* Done poorly, it sounds like an exam, and patients shut down.

The most cited evidence behind the teach-back method comes from Dean Schillinger and colleagues, who audio-recorded primary care visits at a public hospital diabetes clinic in 2003. They found that physicians assessed patient recall and comprehension on only a small fraction of new concepts they introduced. When physicians did use a teach-back style assessment, however, the patients whose understanding was checked had better glycemic control than those whose understanding was assumed. The finding mattered because it linked a communication behavior to a hard biomedical endpoint, A1c, rather than to a satisfaction score. Subsequent work, including a 2015 systematic review by Peter and colleagues, gathered the trials that followed and concluded that teach-back was associated with improvements in disease-specific knowledge, adherence, self-efficacy, and, in several studies, reduced readmissions.

“The question is not a quiz of the patient. It is a check on the clarity of the explanation.”

What makes the teach-back evidence base unusual is how reliably small the effort is in relation to the effect. Most communication interventions require curricular time, simulation suites, or behavioral coaches. Teach-back requires a sentence. The Agency for Healthcare Research and Quality now lists it as a top patient-safety practice. CMS quality measures for transitions of care indirectly reward it. Hospitals running readmission-reduction projects routinely cite it as a cornerstone behavior, and pharmacist-led counseling programs that use teach-back have shown measurable drops in medication errors after discharge.

Three mechanisms explain why teach-back communication moves outcomes. The first is encoding: patients who articulate a plan aloud are far more likely to retain it than patients who only hear it. The second is error correction: clinicians discover misunderstandings while the patient is still in the room, when fixing them costs nothing. The third is relational: the patient experiences a clinician who genuinely cares whether the explanation landed. That last mechanism is where teach-back meets the broader empathy literature. A patient whose understanding has been checked is a patient who has been seen.

The Empathy in Medicine Initiative teaches teach-back as a core skill in its free training modules, not because it is a clever trick but because it is one of the few communication behaviors with both the evidentiary backing and the brevity to survive a real clinic. EIMI ambassadors learn three versions of the question, practice a non-shaming reset for when a patient cannot repeat the plan, and study the Schillinger and Peter trials so they can defend the practice to attendings who dismiss it as soft. Students who want this training can find the modules and apply to the ambassador program at /Apply.

Teach-back fails in three predictable ways. It fails when the clinician phrases the question as a test, which embarrasses the patient and produces compliant-sounding nonsense. It fails when the explanation that preceded it was too long, so the patient can only recall the last clause. And it fails when the clinician does not pause long enough for the patient to actually compose an answer. The fix in each case is small. Reframe the question as a check on the clinician. Cap the explanation at two or three concepts before checking. Wait the full uncomfortable five seconds after asking. Clinicians who do this report that the visit does not, in fact, get longer; the visit gets straighter, because fewer phone calls and portal messages follow it.

For students and trainees who want to build the habit early, the path is unromantic. Practice the closing question on every encounter for a month. Notice the moments when you skip it because you are running behind, and notice whether those are the encounters that generate the most follow-up confusion. The thirty seconds at the end of a visit is not a courtesy. It is, on the available evidence, one of the highest-yield safety behaviors a clinician can build. Readers who want structured practice and a cohort of peers working on the same skill can begin at /Training.

Clinicians who commit to teach-back communication for a few weeks notice a second-order effect: the explanations themselves get better. Knowing that the patient will be asked to repeat the plan changes how the plan is delivered. Sentences shorten. Jargon drops out. Numbers get anchored to concrete behaviors rather than abstract targets. The clinician stops saying we want your A1c under seven and starts saying take this pill with breakfast and we will recheck the number in three months. The teach-back at the end is the forcing function, but the gains accumulate throughout the visit.

Health literacy research consistently shows that roughly a third of adult patients struggle with the kind of dense written and verbal information delivered in a typical clinic visit. Teach-back is a universal precaution against that mismatch. It does not require the clinician to guess which patients have low literacy, which is a notoriously inaccurate guess. It applies the same closing check to every patient and surfaces misunderstandings wherever they happen to live. That universality is why the AHRQ Health Literacy Universal Precautions Toolkit lists teach-back as its anchor practice rather than as a special-case intervention.

A common objection is that teach-back will offend more educated patients who feel they are being talked down to. In practice, the framing the Empathy in Medicine Initiative trains, I want to make sure I explained this clearly, locates the check on the clinician rather than the patient and almost never produces offense. Educated patients, in our experience and in the trials, appreciate the precision. They are also, as it turns out, just as likely as any other group to walk out with the wrong dose if no one checks. The Schillinger 2003 cohort was a public hospital population with high rates of limited health literacy, but the subsequent replications in cardiology, oncology, and post-discharge transitions have all extended the effect across literacy strata.