Picture the physician who signs off on a nurse practitioner’s charts. In a lot of the organizations I work with, that physician is in a different state than the NP, has never met them, has never watched them see a patient, and knows them entirely through a queue of charts to review.
This is normal now. A large and growing share of clinical oversight runs this way, across distance and through the record, between two people who may never share a room. Telehealth made it ordinary. The math of the clinician shortage made it necessary.
At its best, chart review catches something before it happens again. But when the same correction has to be given to clinician after clinician, the system is identifying problems without becoming less likely to produce them. Last month I wrote that the supervisory layer is where clinicians either grow or quietly drift, and that the data to tell which is happening mostly sits unused. This month is a harder version of the same problem. Even when you can see who’s practicing well, getting anyone else to learn from them is the part that breaks. Distance is a lot of what breaks it.
There’s an old idea about where the answer to a problem tends to live. In 1990 Save the Children sent Jerry Sternin to Vietnam to do something about child malnutrition and gave him six months. Nearly two-thirds of the children were malnourished, and he had no money for food programs. So he asked a different question. In the same villages, under the same poverty, were there poor families whose children were healthy anyway? There were.
He found what those mothers did differently, feeding smaller amounts more often and adding foods others threw out, like sweet potato greens and the small shrimp in the paddies. Then he built a way for other mothers to practice those behaviors until they held. Within a couple of years, most of the children in the program were no longer malnourished.
The approach is called positive deviance, and its premise is almost stubborn: the solution is already living inside the group, in people facing the same constraints as everyone else.
Health care has used it. At Yale, Elizabeth Bradley and Leslie Curry built a version for quality improvement. In heart attack care, researchers identified the specific things the fastest hospitals did, like activating the cath lab while the patient was still on the way in, and found that faster hospitals had adopted more of those moves. In 2006, a group of US hospitals ran a positive-deviance project on MRSA.
The prevention strategies were no secret: hand hygiene, isolating colonized patients, surveillance. Teams had taught and measured them for years and the numbers still would not move. What changed the results was not new knowledge. It was who found the practice and how it traveled.
What the room was doing
Look at where every one of those stories happened. Sternin’s mothers watched each other cook. The MRSA nurses shared a unit. The fast hospitals worked the same floors. Proximity wasn’t the mechanism.
The mechanism was a group finding and adopting its own best practice. But proximity was the medium, and it did quiet work nobody put on the org chart. It surfaced the strong performer for free, in the break room and the curbside and the “wait, how did you handle that one?” And it gave the group a hallway to carry a practice down once they’d found it.
That hallway moved knowledge in more than one direction. A supervisor could coach a clinician. One clinician could borrow a move from a peer. A team could notice what was working and make it normal. The point was never simply to identify who was best. It was to make what they knew available to the group.
Take the room away and the requirement doesn’t leave with it. A distributed group still has to find its own good practice and still has to adopt it. It just has no hallway to do either in. And the loss compounds, because in distributed oversight the chart isn’t the main window into a clinician; it is usually the only one. A physician reviewing charts from three states away can’t catch the strong NP in the hallway, because there is no hallway.
That makes a caveat I’d flag anyway matter more here: a chart shows documented practice, not the care itself. The clinician whose charts read cleanest is a signal worth having, and it isn’t the same as knowing who is best. Distance leaves you with the signal and takes away the other ways you used to check it.
There’s research behind why this bites. The feedback literature I mentioned last month says less about whether feedback works than about when it works better, and one of its steadiest findings is that the effect is larger when feedback comes from a supervisor or peer close to the work rather than an outside body. Distance is the loss of that closeness. The person best positioned to help is often the one who has never met you.
What actually has to travel
Nobody learns from the clinician two states away — but what if they could? What would it take to change that? You would have to rebuild, on purpose and through the record, the two things the room used to do for free: surface a strong practice and carry it to the next person.
Imagine a review shows that one NP consistently documents not only what they ruled out, but what would cause them to reconsider the diagnosis or escalate the patient’s care. The useful insight is not that this clinician received the highest score. It is the move itself: making the threshold for reassessment visible. That is something another clinician can understand, test, and adapt. A rank cannot travel in the same way, though a practice can.
But translating performance into a teachable move is only half the work. How the practice arrives matters. Being told to work like the person three states over can land as a judgment on how you’ve been working. A benchmark says someone else is better. A useful learning experience says: here is a move that worked under constraints like yours, here is why it may help, and here is room to decide how it fits your judgment and your patients. People rarely adopt a method that arrives as a verdict.
A teaching hospital without a building
This is the corner I work on at Zivian. The product I help build, Elevate, runs the chart review these organizations already owe and turns it into a read on where each NP and PA is strong. That’s the surfacing, the thing the hallway used to do without anyone noticing, and it is the part software is good at.
The carrying is harder, and it is the part we’re building now, iterating on the feedback loop with the physicians who use it and with our innovation partners. My working view is that rebuilding the hallway requires at least three things. First, the system has to detect patterns worth learning from, not just produce a score. Second, it has to translate those patterns into specific, observable practices. Third, those practices have to arrive through a relationship and an experience that preserve professional judgment rather than turn comparison into a verdict.
The goal is to help a supervising physician see what is already working, make it teachable, and help another clinician try it. That looks less like a dashboard and more like a teaching hospital without a building.
That is my provisional answer to the question proposed in this post. The record can become more than evidence of work already done. It can become a medium through which practice travels, if we design for learning and not only measurement.
For distributed clinical organizations, the question is not only whether required reviews were completed. It is what the organization learned, who else needs that knowledge, and whether the system helped them use it.
Next month: what happens to a good measure once you start managing people by it? Somewhere between noticing performance and rewarding it, the number starts to change the thing it was meant to describe.
— Eden