There is a quiet failure built into the way most consultations end. The clinician finishes explaining the diagnosis, the options, the risks, and the plan. The patient nods, sometimes asks one or two questions, thanks the clinician, and walks out. Within hours, depending on the day, the diagnosis they remember and the plan they remember are not always the ones we discussed.
\nThis isn't anyone's fault. It is a predictable feature of how memory works under stress, layered on top of how much information we routinely ask patients to absorb in a short window. But because it is predictable, it is also addressable. And I think we are overdue, as a profession, to take it more seriously than we have.
\nI want to talk about why patients forget so much after a consultation, what they forget, why it matters, and what a thoughtful response to that problem looks like.
\nStudies of patient recall have been done across primary care, oncology, surgery and emergency medicine, in different countries and across decades. The findings are remarkably stable.
\nPatients forget a large proportion of what is discussed in a consultation. Estimates vary by setting and methodology, but somewhere between 40% and 80% of the information given is forgotten almost immediately, and a meaningful share of what is remembered is remembered incorrectly. The more information we deliver, the smaller the proportion that survives the encounter.
\nThis pattern holds even when patients are intelligent, articulate and motivated. It is not a function of education or attentiveness. It is a function of the cognitive conditions of a clinical consultation, which are unusually hostile to memory.
\nThere are at least four reasons that consistently drive recall failure, and they tend to compound.
\nStress. A specialist consultation is rarely a neutral event. Patients arrive with a worry, sometimes a vague one, sometimes a sharp one. Cortisol is up. Working memory is narrowed. The brain is allocating processing capacity to the threat, not to the lecture. By the time I am explaining the surgical option, the patient may already be thinking about the word \"cancer\" or \"spine\" or \"operation\" and not hearing the next three sentences.
\nDensity. A consultation is densely informative. In thirty or forty minutes I might cover a diagnosis, the underlying anatomy, two or three management options, the risks of each, the timeline, the recovery, the follow-up plan, and red flags to watch for. That is a lot of information, far more than most patients encounter in any other setting in their week.
\nJargon. Clinicians, including me, slip into clinical language without noticing. \"Mild canal stenosis at L4-5 with no significant compression of the cauda equina\" is a clear sentence to me. To most patients, it is mostly noise. They will pluck out one or two words, usually the worrying ones, and miss the rest.
\nNo time to consolidate. Memory consolidation requires processing. In a consultation, the patient has no time to digest one piece of information before the next one arrives. By the end of the consultation, even the bits they \"got\" at the time are at risk of being overwritten by the bits that came after.
\nAdd to this the practical conditions: the patient came alone, or with a family member equally stressed, or after a long drive, or fasting, or in pain. You have a setting designed to defeat recall.
\nRecall failure is not random. It tends to follow patterns.
\nPatients are reasonably good at remembering the diagnosis as a label. They are less good at remembering what the label means. \"Spinal stenosis\" sticks; the explanation of what it is and what it isn't often does not.
\nPatients are reasonably good at remembering whether surgery was discussed. They are less good at remembering the alternatives, the risks, and the threshold for choosing one option over another.
\nPatients are notoriously poor at remembering numerical information. Doses, timelines, percentages and follow-up intervals are forgotten or misremembered at high rates.
\nPatients are also bad at remembering caveats. \"We will watch this for three months and only consider intervention if it gets worse\" frequently survives in the patient's memory as either \"we are doing nothing\" or \"we will operate in three months.\" Both are wrong. Both are consequential.
\nThe information most likely to be forgotten is, unfortunately, a lot of the information that matters most for safe self-management between appointments.
\nIt is tempting to treat patient recall as a satisfaction issue. Patients leave feeling unclear, give a lower NPS score, and the clinic responds with better signage. That is not what is at stake.
\nWhen a patient forgets the diagnosis, they cannot describe it accurately to their GP, to a family member, or to another specialist. Care is harder to coordinate.
\nWhen a patient forgets the plan, they may not turn up for follow-up imaging, may stop a medication early, may double up on a medication, or may miss a referral.
\nWhen a patient forgets the safety-net instructions, they may delay presenting with a red-flag symptom, or they may present unnecessarily for a normal post-operative variation. Either way, the system absorbs the cost.
\nWhen a patient remembers something incorrectly, they may make decisions based on the wrong understanding. I have seen patients defer surgery because they thought it had been postponed when it hadn't, and patients turn up for surgery believing they had agreed to a different procedure. Both are correctable, but both should not happen.
\nThis is a clinical problem with patient-safety implications. We just don't usually frame it that way because the failure mode is invisible. The consultation looked fine.
\nThere is a quieter version of this problem that I think about often.
\nSome patients walk into the consultation with a partner, a child, or a friend. Some bring a notebook. Some are health-literate or have a relative who is. Some return home to a household where someone can help them think through what they were told.
\nOther patients walk in alone. Some come from settings where English isn't the first language. Some are exhausted, in pain, or distressed. Some leave the consultation and have no one to debrief with for hours or days.
\nRecall failure is not distributed evenly. The patients least likely to retain the consultation are often the patients with the smallest support systems and the highest stakes. If we leave information delivery to the patient's memory and to whatever they happen to take home in their head, we are quietly building inequity into the way we communicate.
\nA written summary, clear, concise and in the patient's language, is one of the simplest equity interventions in clinical practice. It is also one of the least consistently delivered.
\nIn my view, a useful post-consultation summary is short, plain and structured. It is not the clinical letter to the GP. It is a different document, written for a different reader.
\nA good summary names the diagnosis in plain language, with a brief explanation of what it means and what it doesn't.
\nIt states the agreed plan in clear terms: what is being done, what is being watched, and what is being deferred.
\nIt lists medications and any changes, with dose and timing and what each medication is for.
\nIt lists tests or imaging the patient has agreed to, with a sense of when and where.
\nIt states the next appointment or review, and what that appointment is for.
\nIt lists clear safety-net instructions: symptoms or changes that should prompt urgent contact, and how to make that contact.
\nIt is written in language the patient can read at home and share with a family member or carer.
\nIt does not pretend to be a substitute for the clinical record. It is the patient's working copy.
\nIf you have ever tried to write five of these on top of an already long clinic day, you know why they are inconsistent. The will is there. The time isn't.
\nThis is one of the places where I think AI can do real good, if it is built with the right boundaries.
\nA consultation, captured as an audio source event and structured into a clinical record, contains everything needed to draft a patient summary. A drafting model can produce a first version of the summary in plain language. A clinician can review it. The patient can leave with something concrete, accurate and approved.
\nThe bar I set on this is not low. The summary cannot be a different version of the truth from the clinical note. It has to be the same clinical content, expressed differently for a different reader. The clinician has to be able to review it before it is shared. The patient should be able to receive it through a channel they actually use. Errors should be correctable.
\nThere are bad ways to do this. An AI summary produced without clinician review, with confident hallucinations about a treatment plan that wasn't agreed, is not a step forward. It is a new kind of safety risk. I am sceptical of any patient-facing AI summary that does not include clinician approval as a structural feature.
\nThere are also good ways to do this. A draft generated from the same source as the clinical note, sharing the same provenance, reviewed and approved by the clinician before release, is a meaningful improvement on the current default, which is \"we hope the patient remembered.\"
\nThis is the design we are building toward with Regenemm Voice. Not because patient summaries are a glamorous AI use case, but because they are one of the most under-served quality and safety problems in specialist care.
\nWhen the patient walks out of the consultation, the clinical work isn't finished. We have asked them to carry, in their head and under stress, the information that determines whether they will manage their condition safely between now and the next appointment. The evidence has been clear for decades that they cannot reliably do this on their own.
\nA good post-consultation summary is not a luxury. It is a quiet piece of the duty of care, and it has been quietly skipped for too long because it is expensive in clinician time. That is exactly the kind of problem AI is well placed to help with, provided we build it with clinician review at the centre, not at the edge.
\nIf we are honest about what patients actually take home from a consultation, the case for thoughtful, reviewed, written summaries becomes very hard to argue against.
\nIf recall failure is something you have seen in your own clinic, and I suspect you have, Regenemm Voice is being designed to draft clear, clinician-reviewed patient summaries from real consultations, in language the patient can use at home.
\nSee how Regenemm Voice supports patient-facing summaries
\nExplore the governed clinical record workflow
\nBrendan O'Brien is Founder of Regenemm Healthcare and a practising neurosurgeon.
","frontmatter":{"title":"Why Patients Forget So Much After a Consultation","summary":"Patients forget much of what is discussed after consultations because stress, density, jargon and lack of consolidation work against recall.","date":"2026-04-22T14:00:00.000Z","category":"CTI Research Series","author":{"frontmatter":{"name":"Dr. Brendan O'Brien","image":{"childImageSharp":{"fixed":{"base64":"data:image/jpeg;base64,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","width":40,"height":40,"src":"/static/bc607af343b0c089694f03ab050ab4b9/f0a3b/dr-brendan-obrien.jpg","srcSet":"/static/bc607af343b0c089694f03ab050ab4b9/f0a3b/dr-brendan-obrien.jpg 1x"}}}},"parent":{"__typename":"File","name":"dr-brendan-obrien"}},"metaImage":{"childImageSharp":{"fluid":{"src":"/static/2c08e9c10cd566741a5c8befe6562ad8/7c2d1/patient-recall-abstract-canvas-titled.png"}}},"image":{"childImageSharp":{"fluid":{"base64":"data:image/png;base64,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","aspectRatio":1.193717277486911,"src":"/static/2c08e9c10cd566741a5c8befe6562ad8/27cf3/patient-recall-abstract-canvas-titled.png","srcSet":"/static/2c08e9c10cd566741a5c8befe6562ad8/447ef/patient-recall-abstract-canvas-titled.png 228w,\n/static/2c08e9c10cd566741a5c8befe6562ad8/acc81/patient-recall-abstract-canvas-titled.png 455w,\n/static/2c08e9c10cd566741a5c8befe6562ad8/27cf3/patient-recall-abstract-canvas-titled.png 910w,\n/static/2c08e9c10cd566741a5c8befe6562ad8/03f8e/patient-recall-abstract-canvas-titled.png 1121w","sizes":"(max-width: 910px) 100vw, 910px"}}}}}}},"pageContext":{"name":"why-patients-forget-so-much-after-a-consultation","previous":"clinician-in-the-loop-ai-is-a-safety-model-not-a-slogan","next":"australian-healthcare-ai-should-be-built-here"}},"staticQueryHashes":["1216909356","185262695","2831923220"]}