The case for information persistence — and the next generation of the EHR.
For most clinicians, the note is the painful pill. It’s the work product you have to create to get paid — the thing standing between the visit and the bill. So it makes sense that nearly every AI scribe on the market is racing to do the same thing: get the note done faster. That’s a real problem, and solving it matters. Payment matters, and clinicians deserve to spend less of their lives typing. I’m glad the industry is attacking it.
But getting the note done faster was never the hard problem. The hard problem is everything that happens after the note is signed.
Every EHR inherited the paper chart
Every electronic health record deployed to date has, in one way or another, made notes a problem. The reason is simple: they never really evolved past the paper charts they replaced. They digitized the filing cabinet instead of rethinking it.
It’s why some practices are still on paper — they look at an EHR and don’t see enough benefit to justify the pain. And they’re not entirely wrong. If you have computerized provider order entry (CPOE) — your meds, your allergies, your orders — you’ve captured most of the tangible, safety-driving benefits of going electronic. Beyond that, what is the typical EHR? A system of tabs and folders. A labs tab. A meds tab. A notes tab. A radiology tab. Nothing about it is organized in a way that actually serves clinical reasoning.
You shouldn’t have to know to look
Here’s the concept I keep coming back to: information persistence.
The idea is that you shouldn’t need to know to look for something in order for it to be available to you. If a patient has had a heart attack — an MI — before, that fact shouldn’t have to be hunted for. It should be apparent the moment you open the chart, plainly and immediately, without a search, without knowing which note from which visit to dig into.
Today, that’s not how it works. The place where that MI should live — the problem list, the medical history — is divorced from the record itself, which is really just a note, or a pile of notes. And a note can be entirely accurate and still be functionally useless beyond the moment of care it was written in.
Think about the cardiologist who does excellent work for a patient in the hospital. If that cardiologist doesn’t also structure the important information so it persists — so the next clinician can see it — they’ve done both the patient and that future clinician a quiet disservice. If the MI isn’t persisted somewhere durable and obvious, somewhere outside the body of a single note, then for all practical purposes it might as well not exist. Someone would have to know to look for that particular note, written at that particular time, about that particular problem. That is not a sustainable way to manage care.
And medicine is only getting more complex
People are living longer. Guidelines multiply. Goal-directed and evidence-based therapies grow more intricate every year. As medicine gets more complex, we need simpler ways to understand what’s appropriate for a given patient — which means we need simpler ways to just see what’s going on with them.
What do clinicians actually do today to cope? They copy forward. They photocopy — electronically — their last note, or a discharge summary, and then add to it. The record grows longer and longer because nobody has time to re-summarize a patient’s entire story at every visit. It’s a rational response to a broken system, but it doesn’t structure anything. The problem list and the medication list are still managed separately from the documentation itself.
Which means even a perfect note can fail the patient. You can write something accurate, beautifully organized, fully structured — and if you leave the practice, or you retire, or you die, no one necessarily knows what’s going on with that patient unless they happen to find your note and happen to know to look for it. The knowledge lived in you, and in a document only you knew to write. It never persisted in the system.
What the next EHR has to be
We need better records. We need systems built around the patient — around the topics the patient actually wants to address, and the medical problems that are relevant to them, whether short-lived or lifelong.
Yes, you still need to generate billing artifacts, and those need to be set in stone, because they’re part of the medical-legal record. But underneath that, there has to be a persistence layer — a real, properly clinically designed layer built for the reality of medicine today, not the reality of forty years ago, when EMRs first entered the space.
Concretely, that means flipping the order of operations. Instead of organizing the record around the note and bolting structure on afterward, organize it around the patient’s problems and topics, and let the note — and the bill — fall out as byproducts. The information a clinician already gathers in the course of care should persist on its own: structured, visible, and durable across whoever happens to be in the room next. No extra clicks. No second pass. No reliance on one person remembering to write the right note in the right place.
That’s the bar the next generation of EHRs has to clear. Not a faster note — a record that still makes sense ten clinicians and ten years from now. It matters most for the private practices and independent clinicians who hold so much of care delivery together, and who have the least patience for software that works around them instead of for them.
The note is solved, or it soon will be. The chart is the real work.
Disclosure: building toward this is what I spend my time on. I’ve kept the argument product-agnostic on purpose — the idea should stand on its own.