A clinician has worked at the same facility for years. They take the late add-on or the late admission or the late shift. They come in when the call schedule falls apart. The nurses know which of them to call when a patient starts going the wrong way. Then something happens, a case or a complaint, and a review begins.

No one at the facility calls them. They hear about it secondhand. Action is taken before anyone asks for their account. The people who could have helped are not in the room, and some of them should have been. Not long after, the clinician quietly leaves for another facility. Even if someone tries to keep them, it is usually too late.

I should be clear where I sit. I am the chief clinical officer for an anesthesia company whose clinicians work in about 150 hospitals and surgery centers that we do not own. When a facility handles a review badly, we often lose the clinician, so I have a stake in how this goes. Before this seat, I was a chief medical officer at a facility, and I have seen this from that side of the table too.

Years against an instant

Years of good shifts do not register the way one event does.

Paul Slovic, who studies how people judge risk, put it plainly in 1993: “Trust is fragile. It is typically created rather slowly, but it can be destroyed in an instant—by a single mishap or mistake.” He asked 103 college students to rate 45 hypothetical news events about a nuclear plant in their community. Half of them gave a report that people living near the plant were less healthy than average a 6 or 7 on a seven-point scale of impact on trust. Fewer than one in five rated the matching good report that high.

His explanation fits a hospital as well as a power plant. Bad events are visible. Good ones, he wrote, “more often are fuzzy or indistinct.” A decade of covering the call schedule is fuzzy. A review has a date, a file and a committee.

A few weeks ago I wrote about the trust a leader can spend with one careless email (“Effective immediately”). This is the institutional version, and the cost is larger.

It is the how, not the finding

We tend to assume the clinician leaves because of what the review found. In my experience, that is rarely the reason. Many seasoned clinicians can live with a finding suggesting they could have done something differently, even when they disagree with that finding. What they cannot live with is what the handling told them about the organization or their value.

Joel Brockner and Batia Wiesenfeld reviewed 45 independent samples in 1996 and found the same pattern across them. How fairly a decision is made matters more when the outcome goes against the person. When the result is good, people forgive a sloppy process more readily. When it is bad, the process carries far more of the weight.

Each failure in the handling sends its own message.

Not calling the clinician directly says: you were not owed a conversation.

Acting ahead of the required process, when nothing required the hurry, says: we decided before we asked.

Leaving out the people who should have been there says: no one who knows your work was worth including.

Taken together, they answer a question the clinician may have been asking for years, which is whether the commitment ran both ways. The finding judges the case. The handling judges the person.

Why no one calls

It is rarely malice. Sometimes it is pressure to act fast after an event. Sometimes the call belongs to no one. When a facility has a concern about one of our clinicians, its chief medical officer often calls me, or calls the group. In all of those calls, no one has ever told me I was expected to call the clinician myself. The facility might assume the group will. The group assumes the facility already has.

More often than either, the conversation is hard, uncomfortable and messy, so no one has it.

Leaving ends the repair

Slovic found something else that matters here. Distrust, he wrote, “tends to inhibit the kinds of personal contacts and experiences that are necessary to overcome distrust.” A clinician who stops trusting a facility pulls back. They stop offering to stay late. They stop making the extra call. Eventually they stop coming.

Trust can come back. Maurice Schweitzer and colleagues ran 262 people through a repeated trust game in which a partner broke trust early. A steady run of trustworthy actions afterward restored it. A promise to change sped up the early recovery. An apology on its own did not help. When the broken trust came with deception, it “never fully recovers.”

That was a laboratory game between strangers, not a department, and I would not stretch it further than it goes. The direction still matches what I have seen. Repair takes a series of trustworthy acts, and a series needs the person to still be there. In the pattern I have watched, the attempt rarely comes. When the clinician leaves, the chance for one leaves too.

Our side of the table

It would be convenient to make this an essay about hospitals. When I was a chief medical officer, I watched my own committees do it.

The clinical committees in my current seat are young. They are still learning how to review a case well, and I want them to learn this part before they learn it the hard way. The wrong way. We can act before we call the clinician. We can let a good clinician go without once asking them to stay. We can follow every rule and still tell someone who gave us years that the years did not count.

If you are moving into a leadership seat, you will be the one deciding whether that call gets made.

What I try to hold to is simple, and none of it changes the finding. Call the clinician before you act, unless a patient is at risk in the meantime, or the call itself would put one at risk. Follow the process as written, including the parts that slow you down, every time. Put someone in the room who knows their work. And when it is over, whatever it found, tell them what happened and what comes next.

Most of the clinicians I have watched leave this way did not leave over the finding. They left over the phone call that never came.

Sources: Paul Slovic, “Perceived Risk, Trust, and Democracy,” Risk Analysis 13(6):675–682, 1993; Joel Brockner and Batia M. Wiesenfeld, “An Integrative Framework for Explaining Reactions to Decisions,” Psychological Bulletin 120(2):189–208, 1996; Maurice E. Schweitzer, John C. Hershey and Eric T. Bradlow, “Promises and Lies: Restoring Violated Trust,” Organizational Behavior and Human Decision Processes 101:1–19, 2006.

That second is the subject of this piece. The best safety work in medicine tends to feel like an obstacle, even to the people who built it, and the people who never lived the problem it solved cannot feel it at all.

Three ways to fix the same problem

Most of the fixes we reach for in healthcare fall into one of three tiers. The weakest is broadcast: tell people. Education, a memo, a policy, a sign above the sink. The middle tier is prevent: make the error harder. A checklist, a default, a standard kit, a device that stops the harm as it starts. The strongest is preclude: make the error impossible.

Hot water at a scrub sink is a simple place to see the first two. You can warn people the water runs hot, and some of them will remember. You can install a thermostatic valve that shuts the flow when the temperature reaches a level that burns, and almost all of them will be protected. Safety scientists have ranked these options for decades. The Institute for Safe Medication Practices publishes a hierarchy of effectiveness with six levels, forcing functions at the top and education at the bottom. I use three because three is what I can hold in my head in a committee meeting.

The ladder is not controversial. Ask any group of clinicians which tier is strongest and they will get it right. Ask the same group what they built the last time something went wrong, and the answer is usually a memo.

The strongest fixes are the ones we stop noticing, which is exactly what makes them easy to undo.

Moving up the ladder

Some of the clearest examples of preclusion in medicine started as ordinary injuries.

Needlesticks used to be a routine hazard of working with IV lines. Education did not end them. Sharps containers helped. Needleless connectors removed the needle from much of the work, and federal law in 2000 pushed hospitals to adopt them. The story did not end cleanly. Within a few years, several hospitals reported higher bloodstream infection rates after switching to certain mechanical-valve connectors. Preclusion removed one harm and, in some designs, moved risk somewhere else. That is worth remembering. Preclusion is a design choice like any other, and design choices have side effects.

Concentrated potassium chloride is the example I come back to often, because it involved no device at all. For years, vials of concentrated potassium sat in floor stock beside look-alike vials, and every so often one was administered to a patient undiluted. The Joint Commission warned hospitals about it in the late 1990s, and taking concentrated potassium off general patient care units became one of the first national patient safety goals. No clinician lost any judgment. A vial left the shelf, and with it went the error.

Then there is the connector most of us never think about. For decades the Luer fitting connected nearly everything — IV lines, feeding tubes, epidural catheters. It was a standardization success, and that was the problem. One connector that fits everything is how feeding formula ended up in veins and how drugs meant for a vein reached the spinal fluid. The fix is a family of connectors, ENFit for feeding tubes and NRFit for neuraxial lines, built so that the wrong pieces cannot mate. It is a good fix that has yet to be widely adopted. B. Braun announced what it called the first NRFit conversion at a US hospital in 2024.

Insulin in the anesthesia cart works the same way and is an example of all three tiers with one drug. A vial at 100 units per milliliter and a syringe make a tenfold or hundredfold overdose a matter of one misread line. Taking the vial out of the cart and putting pens there precludes that error at the point of care. Pens come with their own failure, which is why the FDA has warned since 2009 against sharing a pen between patients. One pen, one patient. When a patient needs IV insulin, it is given as an infusion or bolus from a low-concentration preparation, and at many sites pharmacy mixes it. After hours, a vial can still be pulled from a central automated cabinet, which is prevention through distance and access control. Many sites stock premixed drips in that cabinet for exactly that moment, which pushes the IV path most of the way to preclusion. That was the walk the resident made at two in the morning.

The example anesthesia providers take most for granted is medical gas supply. Three systems work together. Cylinders mount to the machine on a pin index that only fits the right gas. Pipeline hoses connect to the machine with a diameter index that only fits the right gas. Wall outlets, in the operating room and in every patient room, accept only the connector for their own gas. Together they make it physically impossible to connect air, nitrous oxide or suction where oxygen belongs. Most of us have never once wondered why.

Even here, preclusion only guards the point where it is built. In 2016, at a hospital in south-western Sydney, an outlet meant to deliver oxygen to newborns delivered nitrous oxide because the pipe behind the wall had been installed wrong and the testing missed it. One baby died. This is why the oxygen analyzer is still on every anesthesia machine. Strong interventions work in layers, not alone.

The option, not the judgment

Here is the objection I hear from clinicians who write protocols, and it is fair. Preclusion works for connectors and gas lines. It does not work for clinical care, because the patient in front of you is never the average patient, and a pathway that removes clinical judgment will eventually hurt someone it was not written for.

That is true about judgment. It is not true about options.

Potassium and insulin are the proof. Neither change took a decision away from anyone at the bedside. Each removed an option that was almost never necessary and occasionally catastrophic. Protocol writers spend most of their effort deciding what clinicians should do. Very few of us ask the other question: which option on the shelf, in the order set or in the kit is almost never right, and could simply be removed? Defaults are prevention. Removal is preclusion. Neither touches the judgment you are trying to protect.

Where our council starts

I chair the clinical council for the anesthesia management company where I’m chief clinical officer, about 150 sites in all. Two weeks ago I wrote about how that council sets standards. Here is the less flattering part. Most of what the council has produced so far is broadcast. Our first guidelines, on medication management and on patient selection, are approved and still live in documents and messages to our sites. They sit on the weakest tier of the ladder.

That is early in the lifecycle, and it is by design. When a group of sites starts with few shared standards, writing one down and sending it out is the place to begin. You cannot build a forcing function around a practice the group has not agreed on.

What moves a standard up the ladder is the team understanding that the ladder exists. That goes for the council and for the clinical teams who carry the work at each site. It also takes the people who own the levers. In most hospitals the pharmacy stocks the cart and the cabinet, and a pharmacy and therapeutics committee decides what is on the formulary, so a council like ours can recommend a removal but rarely makes it alone. A group that does not know the difference between broadcast and preclusion will keep writing better memos. Sometimes the push comes from a failure: the broadcast went out, and the problem came back anyway. When the results of an intervention do not show what we hoped, my favorite question is a simple one. Does the intervention not work, or are we not doing it? A dashboard cannot answer that. Someone has to go and look.

I am aware that an essay explaining the ladder is itself a broadcast.

The cost of the cabinet

The strongest objection to preclusion is the one I felt in the operating room that night. It costs time, at the worst possible moment, and it costs the most to the most experienced people, who are sure they were never going to make the error in the first place. Constraints that feel like friction invite workarounds. I have seen the insulin vial come back, in a scrub pocket, a fanny pack, a bag under the cart. The gas system has the same history. In 2001 the FDA reported seven deaths and fifteen injuries over four years in which a container of the wrong gas had its connector changed to fit an oxygen line, often by someone who did not know the connector was the safeguard. A workaround to a forcing function is often worse than having no constraint at all, because the system now believes the hazard is gone.

I do not think the answer is to give up the constraint. I think it is to build the next layer so that the friction lands where it can be absorbed. The premixed drip in the cabinet exists because someone took the two-in-the-morning case seriously. A preclusion that ignores the emergency will be defeated by it.

The vial

I still remember being annoyed. I also remember that the annoyance passed before the resident got back to the room.

The point of moving the vial was never that I remember why it moved. It is that the resident who walked to the cabinet that night will never have to learn.

His mother, Yaremy Hernández Castillo, is an internist. As Sarah Zhang tells it in a long piece in The Atlantic, she spent most of the next twenty years trying to understand what happened to her son, on her own time, while colleagues told her the theory forming in her head was absurd. She collected fragments as they came. Another mother in Caracas, picking up the same medication for her own son, mentioned a cluster of anesthesia injuries years earlier. An expert reviewing the case said the lesions looked like inherited mitochondrial disease. A genetic panel in Houston found no known mutation, only a short list of variants of unknown significance that drew little attention.

Then in 2020, her cousin’s thirteen-year-old son had knee surgery in Austin and struggled to wake up. His scan showed two spots in the same places. One case in a family can be bad luck. Two is a pattern. A variant on Uli’s old report turned out to run through their shared maternal line, and with researchers in Spain she assembled seven patients. Same rare change in their mitochondrial DNA. Same inhaled anesthetic, sevoflurane, one of the most widely used in the world. The study made the cover of Anesthesiology this June. Professional societies on several continents now recommend asking patients about maternal Venezuelan ancestry before anesthesia.

Zhang has a line about where Yaremy stood years before the answer arrived: “She already held key pieces of the puzzle.”

What slowed her down was that the pieces lived in different places. No hospital that treated those boys ever saw more than one of them. Uli’s anesthesiologist replayed the surgery in his mind and could find no mistake, because inside one operating room there was no mistake to find. The pattern only existed across sites.

I opened our clinical council’s fall retreat with that story last week. I chair the council for the anesthesia management company where I’m chief clinical officer, a group that provides anesthesia at about 150 sites, from academic centers to two-room surgery centers. Most of the people around the table see patients: physicians, chief CRNAs, medical directors. The council exists for a smaller version of the same problem. A group our size can see what no single site can. I’m also the one grading this meeting, and a company like mine has its own reasons to want standards. That is why the council, not management, has to own them.

That raises a harder question, and I had to answer it before the meeting started. Once you can see across 150 sites, how much do you tell them to do?

A frame, not a picture

Yaremy was working a puzzle with one right answer and no box to show the final picture. Setting standards for 150 sites is closer to a LEGO set. Anyone who has built a large LEGO set knows the moment. With ninety percent of the pieces in place, there is only one thing the set can become, and it’s printed on the box. Dump the same pieces on the floor, minus the picture on the box, and they can become almost anything.

Much of the standardization I have seen in healthcare runs like the nearly finished set. Someone senior arrives with the complete picture, a protocol or a policy already assembled, and asks the room to agree. The room agrees, or appears to, and the picture ships to sites that had no part in drawing it.

Hand people the picture on the box and they will build the picture on the box.

So, with one exception I’ll come to, I did not bring one. I brought a frame. The council decides. Management, which includes me, supports the work. Every standard the council endorses is a floor: the thing every site does unless there is a good clinical reason not to, with any site free to set a higher bar if it has the resources. When a clinician practices outside the standard, the conversation should come from a colleague, not an executive. Past that, the picture was theirs to draw.

What they drew surprised me. Handed authority to build and set standards for 150 sites, the room’s first instinct was restraint. In a change management session on the second day, the clinicians kept returning to the same guardrails. Don’t be overly prescriptive. Respect how different the sites are. Call it a recommendation where a recommendation will do. Protect the buy-in of the people who will have to carry the work. People who have spent careers on the receiving end of a finished picture know exactly what it costs.

What came out of that session was a deliverable for local facility governance, and it does not announce a standard. It invites each site’s feedback and asks what questions they want the council to take up. Floors go out. What the sites see comes back, and that return trip is how the next pattern gets seen.

The draft that did not pass

Motion after motion carried that day without opposition. We adopted a guideline for patients taking GLP-1 medications that the room reworked from my pre-read, adopted a policy on sterile compounding, and confirmed how we select patients for surgery centers. A unanimous room should make any leader suspicious, and it made me suspicious. Agreement is easy to get when the person chairing the meeting is also a respected or senior person in it. In my feedback after a break, I asked for more active engagement, and said I wanted to remove any sense, my own included, that I was driving the conversation toward a specific endpoint.

The clearest evidence that this room was collaborating rather than deferring came earlier that day, from the one finished picture I had brought.

I presented a perioperative glucose standard, merged from two existing protocols and aligned to national guidance, with a treatment threshold and a recheck cycle. The room took it apart. Does every operating room have a glucometer? What happens to the schedule when you recheck every thirty minutes? What do you do with the patient who has never needed insulin and whose sugar climbs after the steroid we gave to prevent nausea? Will the surgeons follow it? The point that ended the discussion came from across the table: a written standard we don’t follow is worse than no standard at all.

That was right. We did not adopt it. We formed a work group with a ninety-day clock, and I’m co-leading it with a CRNA colleague.

Had the room deferred to the picture I brought, we would have shipped a policy that read well and failed at the bedside. Instead, the people who would have to live with it showed me which pieces didn’t fit, before anyone had snapped them into place.

What an empty box costs

The strongest objection to all of this comes from the operating room itself. When a patient is crashing, you don’t convene a council. Someone takes charge, says what happens next, and the room executes. We train for that, and we are right to. Some decisions need the picture on the box, and they need it now.

There is a slower version of the same objection, and the council named it themselves. Without a picture, a room can talk without deciding. When the group mapped where change breaks down, missed deliverables for lack of a decision made their list. A floor can also harden into a ceiling if the council never comes back to raise it. And a glucose work group could easily become the place a hard question goes to wait.

My answer is that the frame has to carry a clock. Withholding the picture is different from withholding a decision. The glucose group reports in January. The guidelines that did not reach a vote move by email within a month or two rather than waiting a quarter. A council that never finishes anything has stopped collaborating and started stalling, and the clinicians at 150 sites will be able to tell the difference faster than I can.

Holding the picture back

Yaremy never had the box. She had a scan, a rumor, a line on a genetic report, and eventually a second boy with the same two spots. The answer came from putting pieces from different places on the same table.

Leaders in medicine are trained to do the opposite. We arrive with the plan, and most of the time that is what the moment needs. But when the work is setting how hundreds of clinicians will care for patients at sites I might never walk into, I’d argue the most useful thing I can do is set the frame, keep the clock, and hold the picture back long enough for the people closest to the work to see what the pieces can become.

They built something better than anything I would have drawn.

The response to the email was silence.

The response in person was different. A few members of the department pushed back. A few made mild fun of it. In the end the change took hold, and the group accepted it reasonably well. What stayed behind were occasional comments about my tone, and one or two about my lack of humility.

I had been right about the process and careless about everything around it. It took me longer than I would like to admit to understand that those were two separate accounts.

Leadership is not one dial

We tend to talk about leaders on a single scale, from poor to great, with good somewhere in the middle. James Bailey, who teaches leadership at George Washington University, took that scale apart in a 2016 piece for Harvard Business Review. He disputes “the stubborn resolve that great and good are points along the same stream.” Great, in his telling, “is a force.” It moves people and organizations. Good is about “protecting and advancing widely accepted principles,” the direction that force is pointed. A leader can have either one without the other.

Draw the two as axes and you get four kinds of leader. Bailey calls the leader with both vital. The leader with principles and no force is amiable. The leader with neither is vacant. The leader with force and no principles to steer it is maleficent, “meaning capable of causing harm.”

The labels matter less than the claim underneath them. If force and principle are separate, then more of one does not make up for less of the other. A surplus in one account does not pay down a deficit in the other.

Trusted to deliver, not trusted to lead

One of the clearest examples I have seen was a surgeon who chaired his department. He was very skilled, and everyone trusted his clinical decisions. He was also command-and-control in nearly everything he did, and distant from the members of his department and from other leaders across the organization. For a long time, that was enough.

Then the organization needed significant change across its clinical enterprise, and the department needed someone to carry its voice into that work. It was not going to be him. His colleagues trusted his hands. They did not trust him to represent them, or to bring the change back to them in a way they would follow. Eventually, the chair seat passed to someone else.

Heidi Gardner, who studies collaboration among professionals at Harvard Law School, gives this split a name. She separates competence trust, faith in someone’s “professionalism, skill set, and capabilities,” from interpersonal trust, which turns on “character and intentions.” Her description of having the first without the second is close to what I watched. “Even if you’re the only person who can help me, if I really don’t trust you interpersonally, I’ll basically keep you in a box while we’re solving this problem.”

He had a surplus of the first kind of trust. It did not buy him any of the second.

A behavior, not a person

It would be easy to stop there and file the surgeon and my younger self in the same box. Kim Scott would object, and I think she would be right.

Scott’s Radical Candor grid uses two different axes, caring personally and challenging directly. Challenge without visible care is what she calls obnoxious aggression: “You challenge, but you don’t show you care.” That describes my email well. Her own guidance on the grid is blunt, though. These are “behaviors everyone slips into sometimes, not personality types.” And: “Please don’t write anyone’s name in these boxes.”

I now sit on the other side of that email. Every so often, someone who reports to me sends their own version of effective immediately to their team. I usually agree with the decision. My response is almost always a groan, and a note to coach them on the style rather than the content.

One tenet of coaching I hold onto is that people are creative, resourceful, and whole, and that a good question unlocks what they already know. So the coaching is almost always some version of the same questions. Have you received any feedback or response to your message from the other day? Then, regardless of the answer: have you re-read it since you sent it? If not, take a quick look now. If so, what might you have done differently, in hindsight, to better deliver an important message?

Those questions mark the difference between the two stories. I landed in the wrong box for one email, people told me, and eventually I heard them. The chair stayed in his for years. Whether anyone told him, I do not know. For me the box was a moment. For him it became a description.

Where the maps disagree

These three maps do not line up neatly, and it is worth exploring where they don’t. Bailey’s amiable leader is principled but lacks force, “friendly and pleasant to be sure, but not necessarily animating.” That is a different failure from Scott’s ruinous empathy, which is care that withholds the challenge. And Gardner’s distinction is not about the leader’s behavior at all. It is about what the people around the leader decide.

Bailey’s map also lets my email off the hook. The decision served a sound principle, and the email certainly had force. By his test it was vital. The cost showed up only on the other two maps, in the care Scott asks for and the trust Gardner describes. That is the reason to read all three.

Three maps, drawn for this essay. To see each framework as its author published it: Bailey’s original in Harvard Business Review (HBR limits free articles), Kim Scott’s Radical Candor grid, and Gardner and Matviak on competence and interpersonal trust.

I’d argue the disagreement is useful. Bailey describes what the leader brings. Scott describes how the leader behaves in a single conversation. Gardner describes what the people around the leader conclude after enough of those conversations. They are looking at different sides of the same relationship, and they draw their boxes in different places. The one thing all three agree on is that the axes are separate. None of them lets you trade one for the other.

Why strong clinicians end up there

The strongest objection to all of this comes from the operating room. When a patient is crashing, force without consensus is correct. We train for it. A clear voice saying what happens next, right now, protects the patient, and a leader who stops to build consensus in that moment is doing harm. Effective immediately is sometimes exactly what a clinical decision requires.

I agree with all of that. It is also why so many of us land in the wrong box in other circumstances. The operating room rewards force and competence trust, and so do most of the ways we pick clinical leaders.

Interpersonal trust is clinical work too, and in anesthesia it is the entire challenge of meeting a patient. Patients have consults and pre-op visits to build a trusting relationship with their surgeon, and time to decide whether they trust that relationship before an elective surgery. They usually meet their anesthesia care team on the day of surgery, often for only a few minutes, immediately before heading to the OR. We get minutes to do the same thing, often while the patient is at their most stressed.

We choose the clinician everyone wants on the hard case. Then we ask them to lead a group of peers who do not report to them and do not have to follow them, where the work runs on principle and interpersonal trust, built over months instead of minutes.

Later this week I will spend two days with the clinical council I chair. Most of the people in that room were chosen, at least in part, for how good they are clinically. I was too. That alone will not carry the conversations we need to have. What carries them is whether the people at the table trust each other’s intentions as much as they trust each other’s hands, and whether any of us can say “here is what I think we should do” without saying effective immediately.

Two ledgers

I still believe in the change I made all those years ago. The process was better for it. But the accounts never merged. Being right about the process did not earn me any trust as a person, and the comments about humility were the bill.

If you lead clinicians, or are about to, look at both ledgers. Ask what you are trusted to deliver and what you are trusted to be. Then, before you send the next one, look for the words effective immediately and ask whether the moment requires them.

Sources: James R. Bailey, “The Difference Between Good Leaders and Great Ones,” Harvard Business Review, September 2016; Heidi K. Gardner and Ivan Matviak, “Implementing a Smart Collaboration Strategy, Part 1,” Harvard Law School Center on the Legal Profession, 2020, and interview; Kim Scott, Radical Candor, “Our Approach”.

They are right, and it bears repeating. Adjusted for the rising cost of running a practice, Medicare’s payment to physicians fell by about a third between 2001 and 2025. A practice that loses money every year does not stay independent because a statute says it should.

I think that argument is sound. I also think it stops one room short.

I am the chief clinical officer of an anesthesia management company backed by private equity. I also hold equity in the company. As introduced, this bill could require companies structured like mine to restructure or divest. That is a stake, and you should weigh everything here with it in mind. In my view, ownership is where this problem shows up, not where it starts. It starts as an economics problem and a governance problem, and the bill reaches neither at its root. It leaves the payment problem where it found it, and it answers the governance question only for the practices it covers.

The concern behind the bill deserves its full weight first. Clinicians have watched decisions about their work move further from the bedside, and some of those moves followed a change in ownership. Those experiences are real. The strongest version of the concern is that some owners add a return requirement on top of the budget, and the staffing plan is where it gets paid. Where I would push is on what we are measuring when we decide an owner is the problem.

What the Stop Corporate Takeovers of Physicians Act regulates

The headline is ownership. A practice would have to be majority owned and controlled by licensed practitioners. The more interesting part is further down, in the limits on management companies. A management company could not control hiring, work schedules, compensation, staffing, credentialing, clinical standards, coding, pricing or payer contracts, among twelve items in total. As introduced, it would take effect in a year, and existing arrangements that do not comply would be required to divest.

The proposed list of limitations is, in reality, a list of governance decisions. Who decides how many people are in the building. Who decides what they are paid. Who decides who gets hired. Who decides the clinical standard. Most of them are decisions patients feel. The bill’s answer to my first question below is a licensed practitioner in control.

Then the bill makes the answer depend on who owns the practice. Hospitals and nonprofits are exempt from the ownership limits. The exemption covers any hospital, including shareholder-owned ones, and “hospital-affiliated clinics,” a term the bill never defines. If the concern is a return requirement sitting on top of the staffing plan, that mechanism can exist under any owner, and the exemption does not ask whether it does. A hospital-employed group would still have someone deciding staffing levels, compensation and hiring. So would a physician partnership. So does every practice in the country. The decisions stay when the owner changes. What the bill changes is who must hold them, and only in the practices it covers.

Three questions for any owner

I sit where those decisions meet the clinical work. In anesthesia they are unusually concrete, because what we provide is coverage. A room runs or it does not. Someone decides how many rooms run after three in the afternoon, how many rooms each clinician covers, and what happens when a clinician says a patient is not ready for the case on the schedule.

Each of those decisions has an owner. Not a shareholder. A person. The ownership debate rarely asks who that person is.

So here is the test I would apply to any practice, whoever holds the equity. A private equity platform. A health system. A partnership of physicians who have practiced together for thirty years. It is also the test I would run before taking a leadership role inside any of them.

1. Who can overrule a clinical staffing decision, and does that person hold a license? Every organization has a budget, and every budget eventually meets a staffing plan. That is where the economics problem lands. When payment falls in real terms, the difference is absorbed somewhere, and under any owner the staffing plan is the likeliest place. The question is what happens when a clinician says the plan is unsafe. If the escalation path ends with someone who has never been responsible for a patient in that room, that is the problem to fix, whoever owns the practice.

2. Where is the clinical standard written, and who can change it? A standard that lives in one experienced physician’s head leaves when they do. A standard that is written down and measured by clinicians can be defended to a hospital, a board or a buyer. A standard that is never measured is a preference.

3. What happens to the first two answers when the owner’s horizon arrives? Every owner has one. A fund has a hold period. A health system has an operating budget and a margin target. A physician partnership has a founding generation that eventually wants to retire. They are not equal. A fund’s horizon is the only one with a date on it. That makes it the most predictable risk, and the easiest to ask about in advance. The horizon is not the problem. The problem is when the first two answers change because the horizon moved closer, and the clinicians and patients carry that risk without being asked.

Holding my own company to it

I owe these answers too. I will not grade my own organization (or any, for that matter) in public, so hold me to the questions instead. They are the same ones I would want a clinician to ask before joining us, and the same ones I would want a hospital to ask before signing with us.

What patients meet

A patient on the morning of surgery does not meet a cap table. They meet the clinician at the bedside. They also meet the consequences of whoever decided how many rooms that clinician was covering, what standard they were practicing to, and whether anyone was measuring it.

Congress may change who is allowed to own the practice. Whether or not it does, ask who is allowed to govern it.

Sources: AMA, Medicare physician pay since 2001; H.R. 10444, bill text; Stop Corporate Takeovers of Physicians Act, one-pager, introduced September 16, 2026.

Then the central line infection rate went up.

The care had not gotten worse. It had gotten better. That rate, CLABSI in the reports, is infections per thousand line days. The patients who still had lines were the ones who could not do without them. They were sicker, they stayed longer, and so did their lines. Their infections stayed in the numerator. The lower-risk line days, the ones that rarely turned into an infection, left the denominator. What remained was divided by far fewer days.

This is common. It is common enough that infection prevention researchers have made the case for a metric that accounts for how many patients have lines at all, not only for what happens once they do, because reducing unnecessary device use may select a higher-risk population that current risk adjustment does not account for. Their phrase for what follows is “a paradoxical increase” in the standardized infection ratio “for facilities that may be high performers.” That ratio is the one a hospital is publicly reported and paid on.

Those numbers moved without anything about the care getting worse. Every quality number a clinical leader will be handed can do the same thing, in one direction or the other.

I have a stake in this argument. I help lead a company that is judged on numbers like these, and when one of ours goes the wrong way, “the counting changed” is the first explanation I reach for. Some of ours are in play right now. I will not say which, or which way, so hold me to the test at the end of this essay. It is the one I have to pass.

Why a CLABSI rate goes up when the care improves

There are three ways the counting moves a number without the clinical care changing. The definition of what counts changes. How hard we look, and who does the judging, changes. Or the population left in the denominator changes. The central line story is the third. It is the easiest to miss, because the change that caused it was the improvement itself.

The first two are better documented, and the best-documented example runs in the flattering direction.

When the definition changes

In January 2015 the national surveillance definition for catheter-associated urinary tract infection changed. Yeast stopped counting. So did cultures below a hundred thousand colony-forming units per milliliter. By the middle of that year the national ratio of observed to predicted infections had fallen to 0.55, an estimated forty-five percent reduction in a single year.

CDC did not call that progress. It told hospitals to adjust their older numbers before claiming a prevention effect, and warned that hospitals “may appear to have met the HHS goal even if they still require further CAUTI prevention efforts.”

In 137 intensive care units measured the year before the change and the year after, the urinary infection rate fell 44.8 percent. The central line infection rate in the same units rose 27 percent. A fungal bloodstream infection that would once have been attributed to the urine culture now had no urinary tract infection to attach to, and in a patient with a central line it was counted against the line. Central line infections from Candida rose 91 percent.

Same units, one year apart. A number that falls gets a slide in the board deck. A number that rises gets a meeting. Both should invite the question of how it was counted, but rarely do.

It is not only CAUTI. Applied to the same 5,804 surgical wounds, one accepted definition found 19.2 percent of them infected and another found 6.8 percent. A definition alone can move a rate almost threefold. Same wounds, same patients, same hospitals.

How hard you look

How hard you look has an equal impact. Across English hospitals, knee replacement infection rates were 4.1 percent where follow-up after discharge was thorough and 1.5 percent where it was not. The authors wrote that hospitals doing the better surveillance will be penalized.

A Swiss program covering 164 hospitals and 187,501 operations found that the share of infections detected only after the patient went home ran from about a fifth after colon surgery to more than nine in ten after knee replacement. A program that stopped looking at discharge would have missed more than nine in ten of the knee infections. The authors titled the paper for the finding: Who Seeks Shall Find. Their conclusion was that intensive surveillance after discharge creates artificial differences between programs. As it turns out, the harder you look, the more you find.

I have seen the same thing in my own specialty, across a number of implementations in different systems over a number of years. When an anesthesia group moves from paper records to an anesthesia information management system, its documented rates of hypotension, hypothermia and difficult airway go up, in every one I have seen. The system records blood pressure and temperature from the monitor on a fixed interval for the whole case, and it asks for the number of intubation attempts and the view grade in fields of their own. A clinician charting by hand in the middle of a case records what there was time to write down, and adverse events are counted off what the clinician reports rather than what is measured against a standard definition. I have compared sites across that line and taken longer than I should have to see how much of the gap was the record. Some of it may be the patients. Until both sites are measured with the same kind of record, there is no way to know how much.

Who calls it an infection

Deciding which cases count sounds clerical. It is not.

When surgeons and infection preventionists in ten countries were given the same twelve written cases, agreement among the surgeons was 0.24 on a scale where 1.0 is perfect and zero is chance. The infection preventionists reached 0.41. After the surgeons read the formal definitions, their agreement fell to 0.09.

Consistency is not accuracy either. In seven Dutch hospitals, the people doing surveillance on colorectal surgery agreed with each other about 95 percent of the time. They still disagreed with an outside panel on roughly one case in eight.

Those judgments become a person’s number. It lands in an evaluation file with a name on it, or in a case review with the clinician who last cared for the patient, and that clinician rarely hears who decided which of their cases counted, or whether that changed.

The test

Before you ask what happened to the patients, ask what happened to the counting.

That is a question with an answer, and the answer is three dates. The last time the definition changed. The last time the way we capture it changed: a new record system, a new reporting form, a new surveillance protocol, a new person or committee deciding which cases count. The last time the population it counts changed: a line-removal campaign, a new discharge pattern, a service line that opened or closed.

Put those three dates on the slide next to any trend line. Better yet, mark these changes on the trend line directly. If one of them falls inside the window the line covers, the line cannot be read as care until someone has separated the counting from the care. If none of them does, the counting stops being an explanation. Infection prevention and quality analytics can usually do that separation if they are asked. Ask for two things: the older period recounted under the current rules, and the denominator plotted beside the rate. They are rarely asked, because a number that improved does not prompt the question and a number that worsened prompts a different one.

When one of our own numbers moves, in either direction, I owe the three dates, and if one falls inside the window, the older period recounted under the current rules, before I credit the care or blame the counting. Without that, the counting is not my explanation, and I do not offer it.

The people who pulled those central lines did the right thing. The number said they had made things worse. Both were true on paper, and only one was true for the patients.

Earlier that day a decision had landed that I did not believe was justified and that was not mine to reverse. I was going to carry it out anyway.

I drove home. I told my wife that I was not okay with this. Five words. I did not elaborate and she did not ask me to, and then I sat down at the dinner table and was no use to anybody in my family for the rest of the night.

That is the entire scene. I want to be careful not to exaggerate it, because the size is not what I am writing about.

I have been in this position in more than one seat and more than one organization. The particulars change more than the shape does.

This newsletter has been about healthcare from inside the rooms where it gets decided — what the systems do, who carries the cost, which questions never get asked in time. Almost none of it has been about the person sitting in the chair. This one is.

Before I go any further, one obvious point. I hold the seat I am about to describe and I am paid to hold it, and the argument I am going to make — that this kind of weight is information rather than weakness — happens to be a convenient argument for the person making it. Read it knowing that.

The hour is the important detail in this scene, and anyone who has held a seat like this one already knows it. The nights you stay until nine or ten are ordinary nights. Those are the nights the work is winning, and there is a satisfaction in them that is easy to mistake for something healthier than it is. Leaving at seven is a different thing entirely. You leave at seven because you have stopped being able to do anything useful, and going home is the only honest option left in front of you.

I have watched other people do this for twenty years without once labeling it as a category. We have a full vocabulary for the clinician who is carrying too much, and we have built conferences and committees and wellness programs around it, most of which have not changed anybody’s Tuesday. We have almost nothing for the person who made the constraining decision, or for the one who was handed it and had to execute it. I had assumed that was because nobody had looked. That turns out to be wrong. I went looking and the work is there, almost all of it about nurses rather than about physicians or executives, which is a different problem and a smaller comfort.

The weight is not the decision itself. I believed it was for a long time, and I suspect most people who take these roles have it wrong in the same way. You imagine the hard part is the choosing. Sometimes it is. But the weight I am describing arrives after the choosing is over, when there is nothing left to decide and nothing left to argue, and you still have to carry the decision out in a body that disagrees with it. It has a name — moral distress — and finding that out was useful to me, because it meant I had been describing something ordinary rather than a private defect. That is all I need it for.

By Wednesday morning it had changed. What changed it was not comfort and it was not resolution. The situation did not improve. The decision stood. Nothing about Tuesday was any better on Wednesday than it had been the night before.

What moved was that I heard where my team actually was. They were the ones who would have to stand in front of their own teams and answer for something they were not permitted to explain, going into a weekend, with nothing anybody could act on. Measured against that, my Tuesday evening was not the hard part of the week. It was the part I had been treating as the subject.

Being petulant and self-centered in that moment was going to make the week worse for everyone.

Weight that has nowhere to go does not stay with the person carrying it. It comes back out as reactivity, and reactivity from an executive chair gets paid for by people who had no part in the decision.

I know what that looks like because I have been the one doing it, and it does not look like temper. It looks like a decision made faster than it deserved, because I wanted it off my desk. It looks like a meeting run short of what the agenda actually needed. It looks like a question I did not ask, because asking it would have opened something I did not have the capacity to manage that day. None of that reads as weight to anyone else in the room. It reads as decisiveness, or efficiency, or trust. And the cost of the question I did not ask lands on somebody else, weeks later, in a form they will never trace back to a Tuesday.

That is a mechanism and not a mood, which is why I distrust the advice to simply sit with it. Sitting with it is what I was doing at seven o’clock. What actually discharged it was being given something real to solve that belonged to somebody other than me.

I still work clinically a few shifts a month. That is not the same job as the one the people I stand next to on those days are doing, and I know it. Part of why I keep doing it is selfish in a way that took me years to understand. Clinical work is the only part of this job where the decision, the cost, and the person it lands on are in the same room within the same hour. Everywhere else in an executive seat there is a delay, and sometimes the delay runs to months. The delay is where the weight collects.

What the delay actually does took me a long time to see. In a room, the consequence arrives while you are still standing in it, and there is closure. You see who paid. You say something to them, or you do not and you at least know what there was to say. Either way it ends somewhere. Push that same sequence out by weeks or months and the closing never comes. You do not meet the person who paid. Most of the time you are never told. So there is nothing for the weight to close against, which is the whole reason it goes looking for somewhere else to land.

I do not think I can take anyone through something I have not been through myself, which means the part of this work that looks like weakness is the part that qualifies me to do it.

The objection to that is the one I take most seriously, and I would raise it myself if I were reading this from the other side. An executive describing what a decision cost him, while holding the authority that produced that cost for somebody else, is a difficult thing to read and it should be. I went looking at the severe end of this too, and there leadership shows up on one side only. Leaders are what harms people. They are not people who get harmed. I am not going to argue with any of it. I was the one in the room. Whatever the week cost me, it cost the person on the other side of it more, and the ordering there is not close.

What I would say is narrower. The distress is not a claim on anyone’s sympathy and it is not a defense of the decision. It is information. It tells me the decision had a real cost, and that I am still close enough to the work to feel it. The day I stop feeling it is the day I should not be trusted with the next one.

On Wednesday morning I made a call before seven, pushed for a different way to carry it out, and agreed on most of it. Later that day I sat down with the team, and instead of opening the way we normally open those, with what happened and what do we do now, I asked them how they were feeling about all that had happened. That is not a question I had ever led with in a debrief. It took about four minutes and it changed what the next hour was able to do.

Then we worked the rest of the day. The decision did not change. Everything downstream of it did.

It goes the same way every time. Something in a report does not sit right. I ask for the number underneath it, and the number does not exist. Not because it would be hard to produce — the data is sitting in a data warehouse somewhere, and the pull is an hour’s work for somebody who knows where things live. What is missing is not the information. It is the decision that the information was worth having.

So I build it myself. A data pull, a pivot table, a chart. I bring it to the meeting and put it on the table, and often enough the conversation in the room changes direction. Then the next period arrives, the same number is missing again, and I build it again.

A number of people have asked me this month about a case that is on the mind of every anesthesia clinician, every hospital pharmacist, and every quality and safety professional I know. It sits now in a publicly posted regulatory finding that a great many of you have already read. I am not going to tell you what happened in it.

That is not modesty, and it is not caution dressed up as principle. It is the standard I hold in the seat I sit in. I lead an anesthesia company and I still work clinically a few shifts a month, and I do not judge the care in any case I have not reviewed in detail — where detail means two things and not one. A full review of the record. And conversations with the people who were actually there. Without both, any judgment I offered would be premature and would almost certainly be wrong on the facts. The record tells you what was documented. The people tell you what was happening. I have often found both to be incomplete by themselves.

I will leave that case to the people doing that work, and write instead about something I have seen from inside rooms I was actually in. If you have been reading the same document I have, you will do your own work with what follows.

Here is what I have come to understand about the way a problem reaches somebody in my role. It does not arrive as an event. It arrives as an event that has already been explained.

By the time a number, an outcome, or a person’s behavior gets to me, somebody closer to it has produced a reason for it, and the reason arrives fused to the thing itself. I am rarely in a position to receive one without the other. And the explanations that reach me are not lies. That is the part I misunderstood for a long time. They are almost always true, offered in good faith, by someone with no motive at all beyond answering the question they were asked.

Two of them come up often enough that I can now hear them coming.

The first is we don’t need it, we make good decisions without it. It cannot be disproven without the very data it is declining to produce, which makes it unfalsifiable by construction. It is not an argument. It is the shape of an argument, and it closes the conversation exactly as well as a real one would.

The second is we’ve never asked that question. That one is a statement about the past offered as a judgment about value. It is entirely accurate and it answers nothing. Nobody has asked. That tells you what has happened, and it tells you nothing whatsoever about whether the question is worth asking now.

The person who explains an anomaly to you almost always owns the function and not the decision, which is why the answer is honest, locally correct, and blind in the one direction that matters. They own the pipe. Somebody else carries the consequence. They are not protecting anything and they are not being evasive — the question simply does not feel urgent from where they sit, because the cost of being wrong about it will land somewhere they will never see it.

And this is where I want to be careful about my own part in this story, because I used to tell it as though the failure were curiosity. It is not. In almost every one of these moments I was not convinced. The answer did not pass the smell test when I heard it, and I accepted it anyway.

Not because I believed it. Because one reporting period and a plausible reason is not enough to spend real capital on. If I push on a service line’s numbers in a room full of people who have just been congratulated on those numbers, and I am wrong, I have burned something I will need later for a fight I have not picked yet. So I bank the doubt. I make a note I will not look at. I wait.

I want to be exact about what that calculus is for, because I have just described it in a paragraph about a service line. It is for a number. It is not for a patient. A doubt about someone on the table gets acted on in the moment, whatever it costs me in the room.

What I am waiting for is the same thing to happen twice. Which means the second instance is not the moment I notice. I noticed the first time. Some of what I noticed turned out to be exactly what I was told it was, and I do not get to count only the times I was right. The second instance is the moment I am finally allowed to act on what I already noticed — the moment a suspicion becomes a position I can hold in a room without being asked to prove more than I can prove.

The best example I have of this was a service line profitability report, and what makes it useful is that nothing about it was wrong.

The attributed profitability per case (the margin the finance model assigned down to each individual case in that line) ran far above every other service line in the organization, and the P&L line item did not explain why. When I pushed, the answer came back quickly and it was correct. That line’s costs were built on a different basis than every other line’s, and the difference was large enough to account for the whole gap. No one was hiding anything. The number was accurate.

But the basis was invisible, and the number kept being repeated. People made capital and recruiting decisions off that profitability figure without knowing what produced it, without knowing that it was not comparable to the figure printed beside it, and without pricing what would happen if the difference underneath it narrowed. Every one of those decisions was locally reasonable. Nobody was ever in a position to ask the aggregate question, which was how much of our strategy was resting on one line’s cost treatment — because each decision only ever touched the number, and never the basis.

That is the thing worth taking from all of this. A wrong number gets caught. A number that is correct and unexplained propagates, and everyone downstream quietly inherits a dependency they never agreed to take on.

The obvious objection is a fair one, and I have had it put to me directly. A leader who chases every anomaly is a leader nobody brings anything to. Investigate on suspicion alone and you will spend your organization’s patience on noise, and the next time something real surfaces the person holding it will decide it is not worth the meeting. Two data points, besides, are still only two data points.

All of that is true. But notice that the second instance is a cheaper trigger than the one most of us actually use, not a broader one. It fires far less often than investigate anything that looks odd, and it fires considerably earlier than investigate after something goes badly enough that nobody can explain it away. Most of us are running the second rule and telling ourselves we are running the first. And two instances are not evidence. I am not claiming they are. What the second one buys is standing — permission to ask a question I could not justify asking the first time. What the two-instance test asks for is not more scrutiny. It is a way to hold a suspicion across a reporting period without either acting on it prematurely or losing it entirely, which is the thing I have never had a good system for.

Which is the part that should embarrass those of us who work on the business side of a hospital, because medicine solved this a long time ago. Morbidity and mortality conference, peer review, event reporting, sentinel event review: every one of them exists so that a single instance lands somewhere other than one person’s memory, and the sentinel event review exists because sometimes one is already enough. They are imperfect and they are gamed and they are sometimes theater, but they are a system, and the whole point of the system is that no one clinician has to carry a suspicion alone across six months and remember to raise it. We built that apparatus for clinical care. On the business side we built it only for the number that misses. Nothing fires when a number is correct and unexplained, and I have spent most of my career in rooms where the equivalent question had no place to go.

Which brings me back to the spreadsheet, and to the part of this I like least.

Every time the information was missing, I built it. I brought it to the room, I made the case, and the decision usually got better for it. What I never did was treat the absence itself as the recurring event. Each one felt like a one-off, because each one produced something — a chart, a number, a decision that went differently. The series produced nothing at all. There was no artifact for the pattern, so the pattern never came up for review.

The loop never got fixed, and the reason it never got fixed is that I kept covering for it. My willingness to build the thing by hand is precisely what kept anyone from having to feel the cost of its absence. I was not exposed to that failure. I was the reason it survived.

So the question I would put to you is not whether you are curious enough. I suspect you are. It is narrower, and harder, and you can answer it this week: what is the thing you have now built by hand twice, and who in your organization is comfortable because you did?

I opened by acknowledging the case mix difference. The practice ran more heavily to cancer than the group it was being compared against, and I said so. I said the cases ran longer. I said the patients arrived with more coexisting disease. I said there was more transfusion. I said all of it before the surgeon had to.

Every one of those things was true.

Then I said the number was nuanced.

I have thought about that word for a long time since. It is the most useful word in the executive vocabulary, and the reason it is useful is that it concedes a problem without naming one. It sounds like candor. It cannot be checked. It let me be honest and unhelpful at the same time, and the person across the table walked out frustrated and with nothing that could be used to improve.

I have described that meeting since as giving someone the benefit of the doubt. What it actually did was close the subject.

And the case mix was never the surgeon’s missing piece. Anyone operating on that population knows their patients are not the comparison group’s. What the surgeon did not have was the instrument — what a standardized infection ratio even is, let alone what had gone into that one. I had the vocabulary, and I had someone else’s analysis. What I had not done was open the model and read what was inside it, and I was the only person at that table who could have. Arriving with the mitigation already assembled meant the surgeon never had to make the case, and so never received the one thing that would have made the case without me in the room.

What the risk adjustment accounts for, and what it does not

Everything I said in that room was accurate. None of it was usable.

What I did not say was what the number was built from. I came to the table with a value and detailed analysis done by another team. Carrying a number and understanding it are separate activities, and I had done the first while sounding like I had done the second.

The risk adjustment behind that ratio, under the inpatient model in force before it changed, carried five things. Whether the patient had diabetes. An ASA physical status class. Whether body mass index was thirty or above. Age. And a flag for whether the hospital is enrolled as an oncology hospital.

That last one is a fact about the facility. It is not a fact about the patient on the table.

Cancer is not in the risk adjustment model. Not as a diagnosis, not as an indication, not anywhere. The effect is published and it is not small. Gynecologic cancer carries an independent adjusted odds ratio of 1.54 for surgical site infection after hysterectomy across 66,001 of those operations at 166 hospitals, and a pooled 1.49 across 681,695 patients in twenty-three studies.

The model has been rebuilt since that conversation. It now includes nine variables instead of five, including how long the operation took. It is a better and more specific model. Cancer still is not one of the model variables.

So I was right about the case mix, and the instrument could not represent it. Not because anyone decided cancer should not count. Because a hospital enrollment flag was used where a patient variable belonged, and because that flag adjusts every surgeon in an enrolled building identically, which means it cannot tell apart the two practices the number was being used to compare.

Now grade the other things I said in that room, which I did not do at the time.

Comorbidity was already in the number. Diabetes, physical status, body mass index and age are four of the five variables, so I gave credit for something the model had counted — the adjustment had already run in the surgeon’s favor before I opened my mouth. Transfusion has never been a covariate, and does not hold up well in the literature either: null across five randomized trials, and dropping out on multivariable adjustment in the hysterectomy analysis above.

The longer operating times were the exception. Operative duration had no term in the model then in force, which made it the one thing I said that the instrument could not already answer. I said it once and moved on.

So of the three: one already counted, one unsupported, one real and unpressed. None of that was clear to me at the time, which is the point.

A group working through 6,142 procedures at one public health system added clinical variables to the CMS set and improved the model’s discrimination from .675 to .735, calling the current risk assessment overly simplistic. In their own matched comparison the uterine cancer proportions did not differ. I include that second sentence because leaving it out would be the thing this essay is about.

The floor is not the standard

NHSN will not calculate a ratio at all when fewer than one infection is predicted, in order to enforce precision of the estimate and comparisons to national data.

That is a floor. It decides whether a number may be printed.

It is not the standard for judging a person. The rule of thumb statisticians use is ninety percent reliability before drawing a conclusion about an individual. Seventy to eighty is considered acceptable for a group.

No single clinician’s volume comes near it. In a 345-surgeon collaborative, eighty-six percent of the variation between surgeons was measurement noise. 168 cases across three years were required to reach even 0.7. One surgeon in 345 had the volume. That was all complications after colectomy rather than infection alone, and the arithmetic gets worse as the outcome gets rarer. At the hospital level — hospitals, not people — ninety-four cases were needed for the same threshold on superficial infection after colon resection, and about half of hospitals had them.

Most of the distance between any two surgeons in that collaborative was due to chance.

The number I walked into that room with passed the only test anyone had run on it. That test was never built to answer the question I was using it to answer.

The two numbers were the same number

Here is the part I understood last, and it is the part that should have been said first.

Infection prevention could hand me a ratio for that surgeon because it was the same ratio as the facility’s. The reported volume for that procedure, over that period, was effectively one practice’s.

The measure CMS publishes and pays on counts only the abdominal hysterectomies where the patient stays overnight. NHSN calls a procedure outpatient when admission and discharge fall on the same calendar day, and every one of those is excluded from the measure that feeds public reporting. The test is the discharge, not the room: a laparoscopic case in the hospital’s own operating room, home that afternoon, is outside the number. Two thirds of hysterectomies were already being done that way by 2013.

So the denominator was not small at random. What decides whether a hysterectomy stays in the measure is whether the patient stays overnight, and what usually decides that is how large the operation was. The cancer operations were the larger ones. In this practice more of them were open rather than laparoscopic, longer, more involved, and more likely to keep someone in a hospital bed overnight. The simpler cases went home the same afternoon and were never measured.

The reporting rule does not sample hysterectomy. It samples the hard end of hysterectomy, and then scores what it finds with a model that has no term for cancer — and, under the version in force then, no term for the approach or for how long the operation took. The model has since added both of those. It still has not added cancer.

An inclusion rule is a sampling decision, and it is almost never independent of the thing being sampled.

That is the loop. The case mix the instrument could not adjust for is the same case mix that decided who appeared in the instrument at all.

The exclusion is not a rounding error either. At one hospital, reportable abdominal hysterectomies fell from 360 in 2017 to 234 in 2021, as same-day discharge went from three percent of its cases to forty-nine. Predicted infections are the actual denominator of the ratio, and they fell thirty percent across those four years. The hospital’s infection ratio rose forty-two percent for every infection it recorded. In 2021 alone, excluding those cases cut predicted infections forty-seven percent and raised the ratio ninety percent.

Here it shrank until the reported population was effectively one practice. The facility’s number and the person’s number were the same, and a rule about what time the patient went home decided how large it was.

The reliability arithmetic above says a hospital needs ninety-four cases before its own rate means much. That test went unrun here, because on paper this was a facility measure, and facility measures are assumed to be large enough — an assumption built on research done at facility scale. It stopped being true the moment a reporting rule shrank the facility to one practice, and no step in the process is designed to notice that.

And this is the recommended practice

The surveillance manual says the ratio can be calculated for specific surgeons. No caution follows it. The 2022 infection prevention compendium goes further and recommends the practice: audit routinely, give confidential feedback of rates and ratios to individual surgeons and to chiefs, benchmark anonymously among peers. It carries one measurement caution, that the one-predicted-infection rule may be harder to satisfy for small surgical programs. That is a note about whether the number can be produced, not about whether it can tell two clinicians apart.

Everyone involved is doing what the guidance says. That is why nothing corrects it.

The asymmetry underneath is the whole thing.

The payment consequence that got the research belongs to the hospital. Six measures weighted equally, colon and abdominal hysterectomy infections combined into one of them, the worst quartile losing one percent of its Medicare fee-for-service payments. That is exactly the level at which the reliability work was done.

The career consequence belongs to a person. Ongoing professional practice evaluation, at an interval that cannot exceed twelve months, is what carries a measure to an individual file. What it usually produces is not a privilege action but a focused professional practice evaluation — the chart audit, sometimes the required-presence proctoring. Further out, an adverse action lasting longer than thirty days is reportable by statute to the national practitioner data bank, and it stays on the record.

That last one is the ceiling, and reaching for it skips what actually lands. Most measure-triggered findings become a letter in a file, a quiet contraction of block time, a schedule rebuilt without discussion, a quality component in a compensation formula. Those carry no threshold, no appeal and nothing reportable, which makes them harder to argue with rather than easier. None of it is confined to surgeons: privileged advanced practice clinicians run the same evaluation at smaller volumes, which makes the arithmetic worse for them.

The reliability of the data used in that evaluation has never been published. I went looking for it. The rigor followed the money.

The objection

I should name my own interest, and the obvious version is not the important one. I lead the clinical side of an anesthesia company, so I am one of the parties that issues numbers like these. The sharper conflict is that an argument against holding individual clinicians to noisy numbers serves the person who employs clinicians. Discount accordingly. I am not arguing that no number should ever attach to a person, and that objection gets its due below.

The disclosure I actually owe is narrower than either of those. Two weeks ago I wrote that the system depends on someone in the room being willing to say how a number was built, and that this job has no audit and no owner. I was that person in this room. I said nuanced. And I have never had the longer conversation with every clinician these numbers reach. I am not certain I have had it completely with more than a handful. Ever.

The strongest objection is one a reader will find before finding me. Published work shows that adding cancer-specific variables to this kind of model did not change hospital quality rankings. At the hospital level that is frequently true. Volumes are large, rankings are coarse, and refining risk adjustment moves fewer institutions than people expect.

That is a finding about hospitals. The unit here is a person, whose volume is a fraction of an institution’s, and whose consequence is a privilege file rather than a payment adjustment. The paper is not wrong. It answered the question the system bothered to ask, which is the question with money attached to it.

The second objection is that if the measures are this soft we should stop measuring. That does not follow. Measurement is how a group finds a problem it cannot see from inside, and what the reliability finding argues for is a change of method rather than an abandonment: judge cases, not rates, when the rate cannot carry the judgment. When a person is on the other end of a number, the precision of that number is part of the number.

What is actually owed

Not that the number is soft. That sentence has no object, and the room hears a verdict on the whole enterprise. Say what it is soft about. This measure. This missing variable. This direction.

Name where it ran in their favor, because it does, and I have just spent a paragraph doing it months late. A leader who discloses softness only when it exculpates the institution has disclosed a preference rather than a fact.

Then pay the part that can be paid, and be specific about what that is, because a vague obligation is the same failure in a different suit.

The human read already exists on paper. Focused review is a human read. The mechanism is not missing — the trigger fires before anyone has asked whether the number could support it, and by then the clinician has been labeled. So condition the trigger. Before a ratio attaches to a named individual, someone writes down what the model adjusted for, what the denominator was, how that denominator compares to the case counts published work says a judgment about a person requires, and which direction the missing variables push. That document goes in the file next to the number.

None of that requires permission. Before you sign an evaluation packet or forward a scorecard to the person it describes, ask those four questions and write the answers next to the number. Your signature is yours. That sits inside someone’s authority in every organization I have worked in.

And say what still cannot be changed. The number will keep counting. Naming the limit is the difference between disclosure and the performance of disclosure.

What I offered that surgeon was sympathy. What was owed was arithmetic.

I do not know what would have happened if I had done that work and said it. The number would have counted anyway. The evaluation would have arrived on schedule. Whatever that ratio was going to do to that practice, it was going to do.

What would have changed is that the surgeon would have known which parts of it were real, and would not have had to take my word for the rest.

This is that conversation, held late and in the wrong room.

Past the Door is free, and it will stay free.

At some point I will open a paid tier. It will be for readers who want more than one piece a week, and the free posts will not get thinner when it happens. I am not there yet, and I would rather know now than guess.

If this is worth something to you, you can pledge a subscription. Substack holds it. You are not charged unless I turn payments on, and if I never do, nothing happens. What a pledge tells me is whether to build the paid tier at all.

It is a good instrument. It forces a room to be specific about something everyone would rather keep vague, and it produces a written record of names against seats.

Work down the list and you will see something the grid was not designed to show you.

For the non-clinical administrative roles there are regularly two or three names in the first two buckets. Someone is ready today. Someone else is close. The room moves through those seats quickly because the answers are already known, and the answers are already known because somebody has been working on them. The names carry development plans with an owner and a date attached, and you can see them on the page.

Then you reach the clinical roles, and the shape inverts. The room quietly agrees that no one is ready today. There may be one name for a year out, offered with a caveat. And then the long list arrives, and everyone contributes to it easily and warmly, and the people on it are described as strong candidates who could be ready in three to five years if the stars aligned. Those names carry no plans. The cells are blank.

A long list at five years with an empty column at the front is not a pipeline. It is a way of feeling good about not having one.

The same people built both columns. The same instrument, the same afternoon, the same organization. Whatever explains the difference, it cannot be that this organization does not develop its people, because the same grid says otherwise a few roles earlier. It cannot be that healthcare is uniquely hard, because the finance and operations roles are part of the same leadership team.

Some of the people not in the ready-today bucket were asked and said no. They were offered the track and they did not want (or were afraid) to leave the bedside, which is a coherent thing to want and is not a developmental failure. The trouble is that the grid does not distinguish the person who declined from the person nobody approached. It records both as not ready. A room that cannot tell those two apart does not have a pipeline either way. It has a list.

There are real differences between the two columns. The clinical seat is usually part-time work laid on top of a full practice, so developing someone for it costs clinical revenue where the administrative version costs a stretch assignment. The clinical ladder has almost no rungs between practicing and chief, where the administrative one has four or five. And the pool does not transfer sideways — you cannot backfill a cardiology chair from an orthopedics department.

Every one of those is a reason the clinical column needs more machinery than the administrative column. It has less.

The interim from another department

A department chair leaves unexpectedly. The conversation has to happen immediately and it has two parts. Who serves as interim, and then, more slowly, whether there are real internal candidates who would hold up in a national search or whether this is going to be an external hire with an internal courtesy interview.

The interim is almost always the harder question, which surprises people who have not sat in that room. The permanent search has months and a committee and a firm. The interim has about a week. And the room goes quiet, because the ready-today bucket for that seat is empty and everyone present knows it.

So the interim gets named from outside the department. A capable chair from a department that could be argued is adjacent. An assistant dean who the administration trusts.

That happens often enough that it has stopped registering as unusual. An interim chair from another department is a public statement, made in writing, distributed to the faculty, that a department could not put forward a single person prepared to hold its own seat for six months.

There are respectable reasons an interim comes from outside. Neutrality, when the internal candidates will be in the search pool. A policy barring the interim from the permanent job. Someone qualified who was asked and turned it down. What separates those from the other thing is whether the room can name the person who said no. When nobody was ready, there is no name to say.

In a community hospital or a private group the same event wears different clothes. The interim medical director arrives from another service line or from the management company, and there is no grid anywhere that records it happened.

Everyone in that department reads the announcement. Nobody says out loud what it reports.

Then the search committee conversation begins, and someone says that it was hard to find a good candidate for this department last time.

It is said as a fact about the market. What it actually reports is a recurrence — we have stood in this room before, we are standing in it again, and in the years between the two visits nobody was developed. The organization experiences a pipeline failure and files it as a hiring problem. Filed that way, it will happen again, and the next person to say that line will believe they are describing bad luck.

Advice is free and advocacy is not

The clinical side of that grid is not short on mentorship. It is saturated with it. Every academic department, every group, every leadership program I have been near has senior people giving generous, sincere guidance to junior people, and the guidance is good.

The distinction I did not hold clearly enough for most of my career is that a mentor talks with you and a sponsor talks about you, in a room you are not in. Sylvia Ann Hewlett’s work on sponsorship, across Forget a Mentor, Find a Sponsor and The Sponsor Effect, made the difference concrete: a mentor invests time and knowledge, which are renewable. A sponsor invests reputational capital, which is finite and does not come back if you are wrong about the person.

That difference explains the number I find hardest to argue with. Of the people who identify themselves as sponsors, only about a quarter actually advocate for their protégé’s promotion — the defining act, the one thing sponsorship is. The rest are giving advice, taking the lunch, writing the letter, and calling it sponsorship.

It is structural. Organizations that build these programs count pairings, because a pairing is an objective, verifiable statistic that describes the program. Advocacy happens in a closed room with no minutes. Everyone involved behaves well and the mechanism still fails.

I should share my own standing. I lead the clinical side of an anesthesia company. Hospitals pay us for clinical leadership, and the group of medical directors I am responsible for holds the seats this piece is about filling. If the argument prevails, it creates demand for something my company sells. Read it with that attached.

The seniority of the mentor is the hinge. Ibarra, Carter and Silva found that high-potential women had more mentors than their male peers and less senior ones, and were advancing less: overmentored, undersponsored, not moving. Inside medicine that framing has become close to consensus. The refrain in academic medicine now is that trainees are over-mentored and under-sponsored.

A mentor who cannot walk into a succession conversation and put your name on the table is not a sponsor, however good the advice and however much you like them. That is a fact about where they sit, not about their character. Most of us have confused the two for years.

The apparatus measures the half that does not decide anything

The profession’s response to the empty column has been to build leadership programs.

A 2025 review in Leadership in Health Services looked at 96 physician leadership programs. Roughly three-quarters measured learner satisfaction. Two-fifths measured knowledge acquisition. Fewer than one in five measured organizational impact. An earlier systematic review of 45 studies found favorable organizational outcomes documented in six of them.

In that review of 45 studies, three tracked whether participants advanced into higher leadership roles. Three.

And the programs that do count it are the ones that built sponsorship into the design. The AAMC’s Council of Deans fellowship places its fellows alongside sitting deans, and it reports what happened next: ten of thirty-seven became deans. ELAM pairs its fellows with senior sponsors inside their own institutions who are accountable for getting them access to governance and succession conversations, and it follows where its alumnae land.

That is not a controlled trial. It is close to the only place in this literature where anyone measured the outcome that matters, and it happens to be where a sponsor was built in on purpose. Weigh it that way.

The rest of the apparatus is not lazy. Placement happens outside the program, months or years later, in a room the program has no access to and no standing in.

Where sponsorship does happen it happens informally, and that has a consequence someone finally counted. Roughly seven in ten sponsors have a protégé who shares their race or gender. Relationship-driven selection inside a narrow leadership tier reproduces that tier by arithmetic. Nobody has to intend it.

The objection, which is a good one

The strongest response to everything above is that sponsorship is favoritism with a better name.

It is also not mine. When Johns Hopkins researchers interviewed twelve clinical department chairs about how sponsorship worked in their departments, the themes the authors came back with were playing favorites and paying it forward.

So let me concede it. It is favoritism. Senior leaders are going to advantage particular people, and they already do, and no policy is going to stop that.

The question is whether the advantage is deliberate and distributed, or accidental and concentrated. Unstructured sponsorship is favoritism nobody admits to, running quietly on proximity and resemblance, and the seven-in-ten figure is what it looks like when someone finally counts. Naming it makes it auditable. Declining to name it does not make it stop.

And the hesitation is not only about fairness. Putting your name behind someone means that if they fail publicly, you were wrong publicly, and no version of sponsorship removes that. It is supposed to cost something. A recommendation that costs the recommender nothing carries no weight.

What it costs to actually do it

The obligation is not to mentor more. We are already saturated, and the saturation is the problem wearing a friendly face.

The obligation is to spend something.

I have found it useful to hold that as a few concrete tests rather than a philosophy, and I offer them as one version rather than the version. Say a name in a room the person is not in, this quarter, and attach your own credibility to it instead of describing them neutrally. Route the visible work — the board presentation, the system committee, the problem with an audience — to the person you claim to be developing, rather than to whoever it would be easiest to hand it to. Answer the question you were not asked: you did not ask, but the person you should be looking at for this is.

And when you cannot get into the room where the placement happens, say so to the person you are mentoring. I can teach you and I cannot place you, and you need someone who can. That sentence costs the most of any of them, and it is the honest one.

If you are on the other side of this — capable, well mentored, and not moving — the question is whether anyone who speaks for you is in the room where that grid gets filled out. Ask the person who has been advising you, and pay attention to whether the answer is yes, no, or a change of subject.

The grid will come around again next year. Someone will read the clinical roles, and the room will agree that no one is ready today, and the long list will be offered warmly by people who mean it.

The list is not the problem. The list has been the same for years. What no one in that room has to answer is what any of us did, in the twelve months since we last said this out loud, to move one name from the long list to the front of it.

Past the Door is free, and it will stay free.

At some point I will open a paid tier. It will be for readers who want more than one piece a week, and the free posts will not get thinner when it happens. I am not there yet, and I would rather know now than guess.

If this is worth something to you, you can pledge a subscription. Substack holds it. You are not charged unless I turn payments on, and if I never do, nothing happens. What a pledge tells me is whether to build the thing at all.

To be fair, we didn’t like the picture the number painted, and the explanation was sound. The measure was derived from fee-for-service Medicare patient encounters, which were under a fifth of our volume. Medicare Advantage already made up a larger share of our payer mix and was not included in the measure at all. Add Medicaid at roughly fifteen percent and a large commercial payer mix, and most of our encounters were invisible to the number on my slide. Whatever the measure showed about those patients, I said, it could not be read across the rest of the population. I had built the slide. I had written the assessment on it. Our management team had reviewed it in prep and agreed, which is how a position arrives in a committee already settled.

The gap I was pointing at was real, and real enough that CMS has since started closing it. Condition-specific readmission and mortality measures are being rebuilt to include Medicare Advantage patients, beginning with the FY 2027 program year. It was not closed then.

A trustee asked whether the findings held in other patient populations.

He was an attorney. He had no clinical training and he was not disputing the medicine. His question was whether the pattern in that population would show up in other populations if we measured them the same way, and if it did, whether the small number was a limitation or a sample.

I did not have an answer. I am not sure the question fully landed on me in the room, because I had walked in with the matter closed.

We took it to our analytics team afterward. It did not occur to me to ask the clinicians who had cared for those patients. By the time a measure like that reaches a board the cases are years back, and no one reconstructs a cohort from memory. What a clinician recalls is the case that stood out, which is a biased sample of exactly the wrong kind. They built a comparable assessment — same outcome, same logic, risk adjustment re-fit on our own data — that could run on any population, and we ran it on one.

Then, because I did not accept the first result, on another. Then across several payer and clinical groups, until I stopped asking for one more.

I would like to call that rigor. Running a test again because you dislike the answer is rigor only by accident, and I could not tell you today which of those runs was diligence and which was hoping.

The number that measure produced was specific to the population it was built for, exactly as I had said. The pattern it was pointing at showed up about as often in every group we looked at.

Then the rest of the team had to arrive at the same place, which took considerably longer than the analysis did. By that point the position on my slide was something we had been claiming for months or years, long before the board meeting where the question was asked. The finding ran against it.

I had been right about the number and wrong about what it entitled me to conclude. The caveat was accurate. The dismissal it was suggesting was not, and I was the person in that room best equipped to know the difference.

That is what I want to write about, because I do not think it was a lapse. I think it is the ordinary condition of the job.

A board does not fail to check the translation. The translation arrives pre-hardened, and the person who pushes on it is outnumbered by arrangement rather than by argument.

The room, and the room before it

The quality committee meets quarterly. An hour, sometimes ninety minutes, and quality is the second item because the first one is approving the last set of minutes. A lot of boards moved quality to the top of the agenda a decade ago, which was the right change and did not alter a line of what is in the report.

A dashboard goes up. Mortality index, PSI-90 (a composite of patient safety indicators), hospital-acquired conditions, a star rating, a Leapfrog grade. Most of it is green. The quality officer built the packet, the chief medical officer takes the clinical questions, the nursing officer takes the falls and the pressure injuries, and the committee accepts the report and moves on.

That hour is the mechanism by which a governing body discharges an obligation it is surveyed against. The Medicare Conditions of Participation put quality of care on the governing body by name.

But the hour is not where the interpretation gets decided. It gets decided in the prep session, among people with different training and the same incentive, and the committee receives the settled version. When the trustee spoke, he was not questioning a slide. He was breaking a room’s agreement, alone, without the vocabulary, against people who had already agreed.

That is the part I had never looked at squarely, and it is why the hour is not the problem.

What the number is made of

Here is what I know now that I would have said I already knew then.

The mortality index and PSI-90 on that screen are not built from the chart. They are built from the coded discharge abstract: the diagnosis codes, and the flag that says whether a condition was present on admission.

That sounds like a technical detail and it is the whole thing.

A mortality index is observed deaths over expected deaths, and the expected side is assembled from coded comorbidities. A more thorough coding effort raises what the hospital was expected to have and lowers the index. Present-on-admission is the cleanest version of the same effect: whether a pressure injury was flagged as present when the patient arrived decides whether it counts against you at all. One field. No change in care.

So the sophistication of a documentation program predicts measured quality independently of the care delivered. A hospital that invests there looks better without a single clinical practice changing.

That is not fraud and it is not gaming the system. Coders who capture comorbidities accurately are doing real and necessary work, and a record that reflects how sick the patient actually was is a better record. Incomplete coding distorts mortality indices and patient safety indicators downward, exactly as thorough coding lifts them. I have said to more physicians than I can count, document well, you should get credit for the acuity of the care you have already provided.

Adjusting a measure for clinical risk is widely accepted. Adjusting it for social risk is less so. Preventable hospitalizations are strongly associated with patients’ social and cognitive circumstances, and the physician value-based payment measures built on them do not capture that. The same omission runs through the hospital-side measures. A hospital serving a poorer and sicker population is measured as a worse hospital, partly for reasons that belong to the population.

The rating systems disagree with one another. Across 2,384 hospitals studied in 2023, CMS Care Compare and Leapfrog were discordant 70 percent of the time, severely discordant in a quarter of cases. Healthgrades and U.S. News disagreed on orthopedic procedure quality between 48 and 61 percent of the time. They are not measuring the same thing. Leapfrog is a safety composite. The star rating spans mortality, safety, readmissions, experience and timeliness on its own weighting. Neither grade is wrong, which is worse than one of them being wrong. Nothing on the dashboard says so.

Every one of those facts is true. Every one of them is also available as a reason to set a number aside, especially when the number tells a story you don’t like.

The problem with being the person who knows this

A methodological limitation is not an opinion. It is a fact about an instrument, and the person who can state it accurately is doing the room a service.

That same fact is the most efficient way to make an inconvenient number go away, and it is available only to the person with the vocabulary to state it. Nobody else in that committee could have made it stick the way I did. If a trustee had tried, I would have corrected him, and I would have been right on the facts while being wrong about the part that mattered.

That is the asymmetry, and I do not think there is a clean way out of it. The caveat and the dodge are the same sentence. They are told apart by what the speaker does next, and the speaker is the only witness to that.

Which is exactly why the check has to come from somewhere other than the person making the claim, and nothing in the structure requires one.

Evaluating the translation would require the clinical knowledge the translation exists to provide. There is usually a physician on the committee, sometimes chairing it, and they have almost never built or validated a measure. Presence is not a check. Every input to the number is audited hard: the abstraction, the coding, the DRG assignment, the quarterly chart re-abstraction on the measures that carry one, with a payment consequence attached. None of the audits touch what the number was said to mean. There is no audit of the translator.

There is also no literature on it that I could find. The translation has never been studied as a variable, which means what I am describing is built from the incentives and from the rooms I have been in. Weigh it that way.

There is one group that could check the clinical half of it. It is the staff who were there, and no structure carries their answer into the boardroom. By the time the number arrives, the answer may not exist anymore.

The incentives run the wrong direction on both sides of the seat, and the first one is written down. A share of what that executive is paid is tied to the measures on the screen, and the plan that ties it goes to a committee of the same board. Upward, naming the softness means telling a board that reports it has been governing on, several of which you delivered, were thinner than they sounded. Downward it is worse. Chartis found that barely half of physicians report significant trust that their executives are honest and transparent with them, which is a finding about executives before it is a constraint on them, and it is still the condition you speak into. Tell the clinical staff the dashboard is soft and you have handed the most cynical version of the room their loudest argument. They will not repeat the instrument has known error. They will repeat the numbers are theater, and the quality program loses credibility.

There is a venue that sounds like it was built for the harder conversation. Every board quality committee I have sat with holds an executive session, and in every one of them the chief executive and the chief operating officer were in the room alongside the clinical officers. It was not a session without management. It was the same room with the door closed, and it mostly got spent on credentialing and personnel.

So the pressures converge on one behavior. Present the report. Answer what you are asked. Do not volunteer the architecture. And the system contains no penalty for that choice. The green dashboard is the state in which the meeting ends on time and no one asks a hard question.

This is a decision with no owner. The report has an owner, with a department and a production calendar. What has no owner is the interpretation on top of it.

I should share my own standing. I lead the clinical side of an anesthesia company, and hospitals pay us for clinical leadership, which means I am paid to be right about the argument I am making. Discount accordingly. The room I own now is a client facility’s quality committee rather than a health system board’s, and the asymmetry there is identical and smaller.

What the trustee actually did

He did not know more medicine than the people in front of him. He knew his own job, which is testing whether a claim generalizes past the case it was built on, and he applied it to material that was not his.

He also did the one thing the arrangement was not built to absorb: he asked a question that could be answered. Not are you sure, which invites reassurance from the expert and ends there. Would the pattern show up in another population measured the same way — which is checkable, which could not be settled in the room, and which therefore had to leave it.

It indicts the prep session rather than the meeting. The instrument that eventually answered him did not exist because it had never been needed. It took a few weeks to build. It could have been built before the slide, and the reason it was not is that the position on the slide had stopped being a finding and become a thing we said.

The more insidious failure is the one I would watch for. A conclusion gets examined once, at the moment it is formed. After that it gets repeated, and repetition is not review.

There is a reason it runs that way, and it is not laziness. Nothing on a governance calendar schedules a second look at a position already taken. Packets are built forward, and last quarter’s interpretation is this quarter’s starting point.

Underneath that, healthcare leaders have had more data than they can use since the EMR arrived, and most of us still decide by instinct, because instinct has been working well enough. A career of adequate outcomes does not send anyone back to examine the belief underneath them.

What I carried into that committee was an instinct with a methodology attached to it. The limitation was real. It was also the reason I did not have to think about the number again.

Trusting your gut is easier than challenging your beliefs.

What the obligation actually is

It is not better numbers. The numbers will stay imperfect.

The obligation is to change what the room is capable of asking, and then to answer it when it asks.

Report the measure and your confidence in the measure. CMS publishes interval estimates on its mortality and readmission measures and uses them to say whether a hospital differs from the national rate. Control limits do the same on a trend. A number without an error bar is a claim, not a finding.

Name which measures are documentation-sensitive, and which move with your payer mix, before the board reads a rise as clinical improvement rather than after.

Bring one case per meeting the dashboard did not catch, say who caught it, and tell them what the committee did with it. Not as color. As evidence about the instrument. De-identified, or in the session where clinical case review belongs. The patient did not consent to be governance evidence.

Say it out loud when a metric moved because coding changed. Every time. That is the sentence that costs the most and buys the most, and it is one I have swallowed in the moment.

And when you reach for a limitation, say what would have to be true for you to be wrong, before anyone asks. That is the one I did not do. The limitation I named was real, I never stated what evidence would change my mind, and so an accurate sentence did the same as a closed door.

The test has a second half, and I failed that one too. Count how many confirmations you require when the answer is unwelcome, and compare it to how many you required when the answer was the one you brought in. I ran that assessment on population after population. I had run the original on one.

None of that requires a board seat or a title. The same function runs at every level where a number goes up a rung: the service line director signing the dashboard that feeds the packet, the department chair presenting to the medical executive committee, the medical director reviewing quality data with a group. Smaller room, same two pressures, same sentence available.

The objection

The obvious one is that I have written a piece concluding that the person the board most needs has my job, at a company paid to supply it.

Fair. What I can put against it is weaker than I would like. The citation physician executives keep in a back pocket is Goodall, which reported roughly 25 percent higher quality scores at physician-led hospitals among top-ranked institutions. A 2022 analysis in JAMA Network Open went wider, drawing on an American Hospital Association file of 6,162 hospitals, roughly a third of which carried HCAHPS ratings and Leapfrog grades. In multivariable models it found no significant association between a physician chief executive and either HCAHPS ratings or Leapfrog grades. The one thing that survived was patients’ willingness to recommend the hospital.

So the strongest evidence for the claim I have an interest in believing is a finding a much larger sample did not reproduce. It failed in exactly the way my slide did, which I notice. I would rather say it than have you tell me later.

What survives sits at the right level and is not as strong as I hoped. Hospitals whose boards had a quality committee at all — about six in ten when it was surveyed in 2006 — showed lower mortality across six common conditions than those that did not. That is a finding about whether the committee exists. It says nothing about who sits on it, or what they caught.

For the claim I wanted to make for this article — that clinical judgment on that committee is what catches a bad translation — my notes had a citation. A 2015 finding, in a quality and safety journal, from a researcher whose work on hospital governance I have cited for years: hospitals whose board quality committees included members with clinical expertise did better on process and outcomes. Author, journal, year, reference number. Every surface I would normally check.

It got opened this week, because somebody noticed in reviewing this piece that my own notes described the same reference two different ways. It is a different paper by different authors in a different journal, and it is a qualitative study built on ten interviews in the English National Health Service — a system whose board structure and accountability chain are not the ones I have spent this essay describing. Good work, and not evidence for anything I wanted to say. The closest honest summary I can give you is an OECD scan across nineteen countries: the evidence that doctor involvement in hospital governance improves performance is limited.

I am sharing that example consciously rather than pretending it didn’t happen. The citation was correct at every surface I did check and wrong underneath, which is this entire essay arriving in my own footnotes. And I did not catch it. It surfaced the way the number did. Somebody noticed something that did not line up, and asked.

In any case the story above does not argue for putting a clinician in the room. There was one. I built the slide.

It argues for putting someone in the room whose job is to ask whether the claim generalizes, and for that person to be hard to ignore.

The question

He was more right than I was, and he could not have told you why in clinical terms. He did not need to.

If you sit in a seat like mine, or you already run a smaller version of it, you know which caveat you did not raise last quarter. And if you chose the bedside, you know which note you wrote quickly on a night you were carrying too much, and the number was built out of what a coder made of that.

The question I would put to either of you is the one I could not answer in that room. What would it take to show you that you are wrong — and have you built the thing that could?

Out-of-network strategy in this specialty sits near one of two ends. At one end the group has built or rented a mechanism that files every eligible claim as a matter of course. The volume is the strategy, and nobody looks at claims one at a time. At the other end sits a number: a charge threshold, above which a claim is filed individually and below which it is written off. Most groups I have seen are nearer the second end, and that threshold is the entire out-of-network policy.

The threshold gets set in a room. Revenue cycle, the billing vendor or the outside firm that would do the filing, a practice administrator, and one clinician, usually the medical director, usually the person at the table with the least time to be there. They set a number once, and it governs thousands of claims until somebody revisits it.

The number is not really about the merits of any claim. It is about float.

Independent dispute resolution (IDR) is the No Surprises Act process for these claims: each side submits an offer and an outside arbiter picks one. Until June 11, taking a single claim there meant fronting a non-refundable $115 administrative fee plus the arbiter’s own fee, which runs $375 to $800 at the arbiters CMS certifies. The arbiter’s fee comes back if you win, so the group is out five hundred to nine hundred dollars a dispute and the only real question is for how long. In the last quarter of 2024 the median determination took eighty-one business days from the day the case was assigned, which is closer to four calendar months, and the disputed payment itself is not due until thirty days after that. Providers prevail in roughly 85 percent of these in the aggregate, so the expected loss is small. The float was not, and the float is what the threshold is priced against.

Batching was supposed to be the answer to that: a stack of small claims filed as one dispute, one fee covering all of them. Lower the cost per claim and the threshold comes down with it. But the criteria did not fit the way an anesthesia group actually bills, so the threshold stayed where it was.

What falls below that line is revenue the group never collects. It impacts how many sites the group can staff, how many rooms one person covers, and how many call nights get spread across the same number of people.

I lead an anesthesia company. The provision I am going to spend the middle of this piece calling undefined names anesthesiology among the four specialties it was written for, which means I have a stake in whether it ever gets written. Read the rest knowing that.

On August 3 a final rule rewriting that process reached its effective date. The trade press said so, the law firm alerts said so, and the specialty societies had already welcomed the rule back in May. A clinician reading any of it would reasonably conclude that the rule governing that room had arrived.

Almost none of it had.

What arrived on August 3 was three things: a definition of bundled payment arrangements, a set of modifications to what has to be disclosed alongside the qualifying payment amount, the payer’s own benchmark number, and the provisions governing the arbiter’s fee, which set when it is paid and what happens if it is not, rather than what it costs. Useful, narrow, and not the reason anyone was watching. The one change that had already altered the filing arithmetic landed two months earlier, on its own schedule. The administrative fee fell from $115 to $15 per party, for disputes initiated on or after June 11, and the Departments that jointly run this — Treasury, Labor and HHS — reached that date by waiving the delayed-effective-date requirements of both the Administrative Procedure Act and the Congressional Review Act so it could take hold five business days after publication. When the government wants a provision to move, it moves.

The date a rule takes effect and the date it reaches the work are different days, and only one of them gets reported.

Everything else sits in the applicability paragraph, which is where rules keep their real calendar. Some of it carries a date. The revised definition of a batched dispute applies beginning November 1. So do the provisions on continuing negotiations and certain withdrawals. But most of the rule hangs on a different kind of clock, and to see it you need one term first. Every dispute has to open with a negotiation period, a thirty-business-day window in which the two sides are required to try to settle before an arbiter ever sees it. That window is the hook the rest of the rule hangs from. And the operative sentence reads that the amendments “are applicable to disputes with open negotiation periods beginning 90 calendar days after the Departments issue guidance announcing that the functionality supporting these provisions has become available.”

Not ninety days after the rule. Ninety days after the government says the software works.

That sentence governs open negotiation, dispute initiation, arbiter selection, the offer and payment-determination deadlines, the eligibility review, and the batching criteria. The Departments state their own expectation in the preamble: all of that functionality is anticipated twenty-four months after the effective date. That is August 2028. The guidance comes on a rolling basis beginning this summer, and rolling means there is not a date. There is a sequence of them, and the order belongs to the agency.

Four days after the effective date the Departments published an implementation timeline guide, and it adds two things the rule did not specify. It sets the build order: the registry first, then the suspension of resubmissions, then open negotiation, initiation, arbiter selection, withdrawals, and extensions. And it says the Departments expect to implement that functionality in Spring 2027, which sits well inside the preamble's twenty-four months. A build order is not an applicability date. Each provision still waits on its own notice and on the ninety days that run from it, and the eligibility review is absent from the order entirely.

Now the part that matters to those of us named in the rule. Buried in the batching criteria is a provision written for four specialties by name. The regulatory text reads: “For anesthesiology, radiology, pathology, and laboratory qualified IDR items and services, the qualified IDR items and services were furnished to one or more patients and were billed under service codes belonging to the same Category I CPT code range, as specified in guidance published by the Secretary.”

The last clause is the entire provision. The provision that finally lets an anesthesia group batch the way its billing actually works turns on a grouping of codes that the rule does not define and hands to guidance. No such guidance has been issued as of this writing. Category I codes are the ordinary procedure codes, and the rule says range where the CMS fact sheet summarizing it says section, which in CPT taxonomy are not obviously the same thing. The provision has a date and no content.

One thing it does settle. Batching under that provision turns on the code range, not on the credential of the clinician who furnished the case. The payment modifiers that identify who delivered the anesthetic sit on top of the same base code and do not move a claim out of its range, so on its face the provision reaches the case whoever gave the anesthetic.

It acquired the date on August 3, and not from the rule. On the same day the rule took effect, CMS posted a notice announcing that batching functionality would be available November 1, and that all batching provisions would apply to disputes with open negotiation periods beginning on and after that date. That notice is the first of the rolling functionality announcements, and it reaches batching alone; the eligibility review sits in the same sentence of the rule, on the same trigger, and has no announcement of its own. Anyone who had read only the rule, carefully and correctly, would have had to tell their group there was no date for batching. That is what we told ours. Last Monday there was one. It came from a notice on a webpage, and it is now the operative fact.

Three constraints bound what a batch can carry, and a fourth bounds what can follow it. The notice caps a batch at fifty line items. The rule requires that every item was furnished inside a single thirty-business-day service window, and that the batch went through its own thirty-business-day open negotiation period ending within four business days of initiation. Then the notice starts a thirty-business-day cooling-off clock after a determination, during which the filer may not come back at the same party over the same item or service. Three of those windows run thirty business days and none of them is the same window. There is a version of this that is a real improvement for a group whose claims share a code range and a month, and a version that is a filing-window problem. Which one it turns out to be depends on a definition that does not exist yet.

And November 1 is not the date a group can batch. It is the date the clock starts. The notice applies the batching provisions to disputes whose open negotiation period begins on or after November 1. That negotiation window runs thirty business days, and initiation follows within four business days of its close. So the first batch filed under the provision written for anesthesiology is a December event, and the letters that open those negotiations go out in November, against a grouping of codes that does not exist.

The strongest objection to everything above is that I have described delay as though it were injury, and the parties being made to wait are winning.

That objection is correct on the facts, and the facts are not close. In the second half of 2025, providers, facilities and air ambulance services prevailed in roughly 85 percent of payment determinations; plans and issuers prevailed in about 14 percent. The prevailing offer came in above the qualifying payment amount 87 percent of the time. Defaults, where one side never submitted an offer, accounted for 17 percent of determinations, and 90 percent of those went the providers’ way.

And the volume is not a rounding error against expectations. The Departments planned for a process that would take tens of thousands of disputes a year and received 489,000 in the first year alone, fourteen times what they had projected. In the last six months of 2025, disputing parties filed 1.37 million. By the end of January 2026 the cumulative figure was over 5.1 million. Three filers accounted for 38 percent of that half-year’s filings. Ten accounted for two-thirds.

That concentration is worth sitting with for a few minutes, because it is the first of the two ends this piece opened on. Filing at that scale is a business, run by outside firms on providers’ behalf, and it is the end where no threshold exists because none is needed. Payer-side counsel describes the volume problem in exactly those terms: “the submission of large volumes of disputes and offers in excess of billed charges” by “providers and their third-party intermediaries.” The group with a threshold is not one of those firms. It is the party the process was nominally built for, and it is the party the threshold exists to protect from the process.

Of the disputes initiated in that period, eligibility was challenged on 42 percent and 19 percent were found ineligible. A government building screening functionality slowly for a process that drew fourteen times its projected volume in its first year and has grown since is not obviously an injustice. It is a queue.

But hold the two halves of this rule next to each other, because the structure concedes something. The provision that lowers the cost of filing was accelerated by waiver and took effect in June. The provisions that screen what gets filed — the eligibility review, arbiter selection, the offer and determination deadlines — wait on a portal. That is not a conspiracy and I would not argue it as one. But the fee cut did not simply arrive faster. The Departments waived two statutory delay requirements to make it arrive faster, and no waiver can build a portal. The result is the same either way: the cost of filing fell in June, and the review that screens what gets filed has no date.

The concepts in the rule are sound, but the mechanics most of it waits on don’t exist. There is now a functional bottleneck sitting just past a cheaper way to initiate disputes. The container holding the queue is going to need to be very large.

Go back to the room where the number gets set. That threshold was priced against a $115 fee and a determination that took the better part of four months. Since June 11 the fee has been $15. And the determinations have gotten faster — 62 percent of them closed within thirty business days in the second half of 2025, against 37 percent in the first half, which puts the median inside the window. Both inputs to that number moved this year. The number has not.

All of which is the part worth carrying out of a payment-policy fight and into any room where someone announces a date. The effective date is the least useful number in a rule. It is the one that gets into the headline, the summary email, and the slide, and it tells you when the document is law rather than when it changes anything. The date that matters is in the applicability paragraph, and sometimes it is not a date at all but a condition — a guidance document, a portal that has to be built, a go-live that has to be announced — held by the agency rather than the calendar.

So the next time someone brings you a rule with a date on it, ask the second question. Not when does this take effect. When does this reach the work, and who decides.

This piece sits alongside the benchmark is the whole game, which covered how the qualifying payment amount is built. The map that holds all of it lives here.

From the room you can’t see.

The roster is familiar: revenue cycle, pharmacy, ambulatory operations, clinical operations, senior executives. The recommendation to offer the therapy came from the clinical service line leader and arrives with the clinical case already made. What this room decides is something else. Whether the therapy is delivered in a hospital outpatient department that carries 340B benefit, or at another site of service. Occasionally the answer is that the organization will not offer it at all, when it cannot be made to work financially anywhere. And sometimes the question runs the other direction: whether an existing site should be converted into a hospital outpatient department so that it carries the benefit.

The number driving the decision sits in the pharmacy line, labeled 340B benefit. It is quantified partly to understand what the program contributes and partly to understand how much the service line’s profitability depends on it. It stays out of the service line pro forma and out of the financial reporting. The figure shaping where a patient will receive treatment lives in a document separate from the one recording the decision.

I have been in rooms like that one. The people in them are conscientious and the work is ordinary, and the meeting sits on a standing calendar because this decision comes up several times a year, every year, at systems across the country.

That room has always run on a number the government never actually quantified. Three weeks ago it measured it. Medicare had spent most of a decade declining to ask hospitals what they pay for the drugs they buy, and the answer, published on July 7 inside a proposed payment rule written so that few people outside hospital finance would ever get through it, was that drugs acquired through the 340B program cost hospitals 33.4 percent below the average sales price, the benchmark Medicare uses to set what it pays. Drugs acquired outside the program cost 2.7 percent above it. Over that same window, Medicare was reimbursing those same hospitals at the benchmark plus six percent. In round numbers, the hospital buys at sixty-seven cents and is paid a dollar six.

That difference has a name in the trade. It is the spread, and it is worth being clear about something before going further: the spread is not an abuse of the program. The spread is the program. Congress built 340B in 1992 by requiring drug manufacturers, as a condition of participating in Medicaid, to sell certain outpatient drugs to safety-net providers at a deep discount. It attached no appropriation, and it attached no requirement that the resulting margin be traced to a low-income patient or reported to anyone. That last line should sound familiar to anyone who read last week’s piece: the reporting requirement is missing in the same place, for the same institutions.

Two honest qualifications. Thirty-three percent off acquisition is not thirty-three percent to the bottom line, because split-billing software, third-party administrators, contract pharmacy dispensing fees and the compliance staff the program requires all come out of it first. And the six percent is a statutory figure; after sequestration the effective rate lands closer to four and a third. The spread survives both. It is smaller than the headline and it is still the largest reliable margin in most hospital pharmacies.

The argument worth having about that conference room is not the one either industry lobby is making. The pharmaceutical industry’s version is that 340B has become a scam. The hospital associations’ version is that it is working as intended. Both are arguments about the average. Look instead at the distribution, because the two ends of it are not the same room, and almost everything interesting about this program lives in the distance between them.

It helps to clarify the size of the program first. In calendar year 2024, covered entities purchased $81.4 billion in discounted drugs, up 23 percent in a single year, and disproportionate share hospitals, the large facilities that qualify by treating a high proportion of low-income patients, accounted for nearly 79 percent of it. That growth did not come from a sudden expansion of the uninsured population. It came from institutions doing what financially rational institutions do. The Congressional Budget Office found that the number of off-site outpatient clinics registered in the program grew from roughly 6,100 in 2013 to 27,700 in 2021, a great many of them in affluent, commercially insured markets. Contract pharmacy arrangements went from fewer than 1,300 locations in 2010 to roughly 31,600 today, spanning something close to 240,000 separate covered-entity contracts. A contract pharmacy is a retail pharmacy that dispenses 340B-priced drugs on a hospital’s behalf and splits the spread with it, which is why a single hospital can have hundreds of them.

Those numbers describe a peak rather than a trend. Since 2020 a long list of manufacturers has restricted contract pharmacy shipments outright, and the fight has now moved to rebate models and state protection laws, with the federal circuits now split on whether those state laws survive. The restrictions are themselves evidence that the extraction is real enough for manufacturers to have spent six years and considerable money trying to stop it.

Here is why the geography matters. The spread on a drug is proportional to the drug’s price and to what the payer reimburses, which means the program pays best on expensive drugs given to commercially insured patients. Oncology sat at the top of that list for a long time, and the pattern is visible in the data: researchers at the USC Schaeffer Center found Part B oncology spending per Medicare beneficiary running substantially higher at 340B sites than at comparable non-340B sites, with 340B eligibility associated with a roughly 90 percent increase in hematology-oncology claims. Association is not attribution, and site-of-care migration and referral concentration both push in the same direction. But site-of-care migration is not a statistical artifact. It is a decision, made on a standing calendar, by the room described at the top of this piece. The researchers are measuring the residue of thousands of those meetings.

Set that beside what the Government Accountability Office has been reporting for years. In a review of 55 covered entities, a sample GAO itself cautions does not generalize to the program, 30 offered discounts to low-income or uninsured patients at some or all of their contract pharmacies. Just over half. The important part is that the other twenty-five broke no rule, because no rule required them to do otherwise. That is the finding. Not that covered entities broke a promise, but that Congress never asked for one.

Now the other end of the distribution. A twenty-five-bed critical access hospital in a rural county qualifies for 340B on different terms than an urban system does, and depends on it differently. Cost-based Medicare reimbursement sounds like it should cover cost, but it does not because it pays a percentage of allowable costs on Medicare patients only and leaves the rest of the building uncovered. The 340B margin fills part of that gap. And here is the uncomfortable part, the one that complicates my own frame: the rural hospital’s margin comes from the same place the suburban system’s does, from commercially insured patients and retail scripts run through a contract pharmacy. The mechanism is identical at both ends. Only the use of the money differs, and only one end has the scale to build a service line designed around it.

The clearest evidence of how thin the rural end of the spectrum runs sits in a contradiction Congress has left in place since these facilities first became eligible in 2023. Rural Emergency Hospitals, the designation created specifically to keep small rural facilities from closing, are statutorily ineligible for 340B. A hospital that converts in order to survive loses the program in the same transaction. The Rural 340B Access Act, introduced in January of 2025, would fix exactly this. It has sat in committee since.

The loss compounds from there, running in a loop rather than a chain. A thin margin makes an oncology service line hard to sustain; losing the line removes the highest-spread category from the hospital’s 340B book; and that thins the margin that sustains everything else. Between 2014 and 2023, 424 rural hospitals stopped providing chemotherapy services. The wrinkle worth naming is that critical access hospitals, rural referral centers and sole community hospitals are excluded from 340B pricing on orphan-designated drugs, which covers a good deal of modern oncology, so the rural end of this ledger was thinner than the urban one by definition. Set the two ends beside each other and the shape is hard to miss. In one room the question is which new therapy to offer where. In the other, the question was whether chemotherapy could continue at all, and for 424 hospitals the answer was no. What that produces is a patient with a curable cancer driving ninety miles each way, every three weeks, for a cycle that used to be given a few minutes from home.

The strongest objection to all of this is correct as far as it goes. Cross-subsidization is how safety-net hospitals have always worked. Revenue earned on commercially insured patients has funded uncompensated care for as long as there have been hospitals, and demanding a dollar-for-dollar accounting that ties each discounted vial to a specific charity prescription misunderstands how hospital finance operates. The honest question was never whether every 340B dollar buys an uninsured patient’s medicine. It is whether the institution’s capacity to serve underserved patients survives without the program, and for a great many covered entities, particularly rural ones, the margin is the institution. The evidence on the other side is mixed rather than damning. A February preprint, not yet peer reviewed and written by authors with a stated critical posture, found 340B hospitals providing somewhat less charity care as a share of operating expenses than non-340B hospitals, 2.16 percent against 2.82. Two scoping reviews, in the Milbank Quarterly and JAMA Health Forum, reached the same unsatisfying place: real revenue, expanded services, inconsistent evidence that the revenue specifically reaches low-income patients.

There is one more thing worth holding onto, and it is the reason anyone on the provider side, at the bedside or in the budget meeting, can cite the GAO findings without becoming a pharmaceutical talking point. There is no appropriation and no line in a federal budget. The discount comes out of manufacturer revenue. The spread, though, is paid on the claim, which means commercial premiums and Medicare coinsurance fund it, and manufacturers argue the cost reappears in list prices, an argument that deserves a hearing it rarely gets from our side. What remains true is that the first-order incidence sits with an industry that is not the patient in the bed. That distinction changes the weight of the critique considerably, and it is what makes an honest accounting possible from the provider side at all.

Which brings the argument back to the proposed rule, and to the part that will be misread. Medicare is not changing 340B. The discount, the eligibility rules and the ceiling price all live at HRSA and none of them change. What Medicare controls is what it pays, and the proposal is to reimburse 340B-acquired drugs at the benchmark minus 33.4 percent rather than the benchmark plus six, which removes the spread on the Medicare portion of a hospital’s drug volume. Because the change has to be budget neutral by statute, the roughly $4.55 billion it recovers gets handed back to hospitals paid under the outpatient system as an 8.44 percent increase in what Medicare pays for non-drug services. Follow the money and the transfer is the story. It moves away from hospitals in the program and toward hospitals generally, which makes hospitals that were never in 340B the net winners, and which makes procedural and surgical service lines at those hospitals the quiet beneficiaries of a rule everyone will describe as a drug cut.

Being specific about what the rule reaches matters, because the piece would be wrong otherwise. The largest spread sits on commercially insured patients, and Medicare cannot touch that book. A rule aimed at the top of the distribution reaches only the part of it Medicare pays for.

Go back to the conference room, though, because this is where the two threads meet. Medicare has been narrowing the payment advantage of hospital outpatient status for years, one service line at a time. Clinic visits went to a physician-office-equivalent rate in 2019. Drug administration followed, effective this past January, at grandfathered off-campus departments. Imaging without contrast is the one proposed for 2027. Each of those narrows what a hospital earns for the service of delivering an infusion. What survived all of it was the drug line, which is to say the 340B spread, and the drug line is the reason the conversion question keeps coming back to that room. A March court decision made it easier still, vacating HRSA’s requirement that a new outpatient site appear on the Medicare cost report before it could buy at 340B prices. The government has appealed. So the honest way to read this proposal is that Medicare, having spent a decade trimming the reward for delivering care in a hospital outpatient department, is now going after the one part of that decision that still pays.

This is also the second time Medicare has tried the 340B piece. In 2018 it cut 340B drug payment to the benchmark minus 22.5 percent, and in 2022 a unanimous Supreme Court threw the cut out, not because a differential rate was forbidden but because the agency had varied the rate by hospital group without the acquisition-cost survey the statute requires before it may do so. This time the survey came first. That is what the 33.4 percent figure is: the evidentiary record the Court said was missing, built and published before the rate was proposed. Hospitals are also still absorbing the remedy from the first attempt, a reduction to non-drug payments that this same rule proposes to raise from half a percent to three. The legal armor is different now, and hospitals planning around a second Becerra are planning around the wrong case.

Now the part I had wrong when I started writing this, and the part most coverage will likely get wrong too. The rule does not fall evenly, and it does not fall on the bottom of the distribution at all. Critical access hospitals are excluded from the outpatient payment system by regulation and paid on cost, so the cut cannot reach them, and neither can the 8.44 percent offset. Rural sole community hospitals, children’s hospitals and PPS-exempt cancer hospitals were carved out of the 2018 cut, and the current proposal carries those exemptions forward, leaving them at the benchmark plus six while they collect the service-payment increase alongside everyone else. They come out ahead. CMS says in the same section that it may revisit the rural exemption in future rulemaking, and it is taking comment on all three. The protection is real and it is discretionary.

Sit with what that means. The payment system has learned to tell the two rooms apart. It sorts by hospital class, protects the facilities where the program is most clearly working as intended, and lands the cut on the large disproportionate share systems that capture most of the spread. The program itself still makes no such distinction. 340B allocates its benefit in proportion to commercially insured volume and contract pharmacy scale, which is very nearly the inverse of safety-net need, and it has done so for thirty-four years. Medicare figured out in one rulemaking cycle what the statute has never been amended to see.

Whether any of this is a rounding error or a service-line decision at your institution comes down to three numbers your chief financial officer already has: the share of your 340B margin that comes from Medicare rather than commercial payers, the share that comes from oncology, and your non-drug outpatient volume. The first sets your exposure, the second sets your concentration, and the third sets how much of the offset comes back to you. If you are at a hospital paid under the outpatient system with a large Medicare oncology book, you are the target. If you are at a critical access hospital, a children’s hospital or a rural sole community hospital, you are not, and the more useful question is what the increase does for the service lines you run. Comments on the rule close August 31, and a comment from a fifteen-bed hospital describing what its 340B book actually funds will carry weight that a trade association letter cannot.

I keep returning to that survey number, because of what it took to produce it. For thirty years this argument was conducted between advocates trading averages, and the government did not know, with any precision, what hospitals were paying. Now it does. And then, having finally measured the distribution, it priced off the midpoint, one national rate for both rooms. The number confirms what both sides already understood, that the program generates a large and real margin exactly as designed. What it cannot tell you is where that margin went, which hospital kept the lights on with it and which one built a suburban infusion suite with it. That is not an unknowable fact. It is known with precision in the conference room this piece opened in, by everyone sitting at that table, several times a year.

For anyone moving from the clinical side into a room where these decisions get made, that is the lesson worth carrying out of a drug pricing fight. Institutions answer to the incentive in front of them rather than to the intention behind it, and reading the incentive correctly is the first thing the new chair will ask of you. There will be a therapy on the agenda, a clinical case already made, and a line in the pharmacy column that decides where the patient goes. Those are the two rooms. They are running on the same statute, and they should stop being governed as though they were the same room.

This is the fourth piece in a thread that runs from what happens when surgery stops subsidizing the hospital through who owes the readiness payment and last week’s look at how executive pay gets set. The map that holds all of it lives here.

From the room you can’t see.

Subscribe now

I have spent more than two decades moving between the clinical rooms and the administrative ones, and the compensation committee is among the quietest of the closed rooms and one of the most consequential. It rarely produces a scandal or a raised voice. What it produces is a figure that helps decide how much room a budget leaves for everything not strictly required, on floors the people in the room will never work a shift.

I want to be careful about the argument I am making, because it is easy to hear a different one. I am not arguing that a given number is too high. That is the argument everyone already has, and it goes in circles, because it turns on a judgment about one person in one year that reasonable people will never settle. The argument worth having is about the features of the room itself: the ones that keep the number out of view, and the ones that push it upward regardless of who holds the seat. I made the same distinction writing about the anesthesia subsidy hospitals fight over, where the subsidy is the symptom and the case leaving the building is the disease. The number is a symptom. The architecture is the subject.

Start with how the number gets built, because the mechanism is more telling than the result. The consultant assembles a peer group, a set of supposedly comparable organizations, and the board sets a target within it: pay the executive at the median of that group or somewhere above it. On its face this is prudent and defensible. It is also, when every board does the same at once, a machine for escalation. If most organizations aim to pay at or above their peer median, and pay rarely moves down once set, the median itself has to climb every cycle, because each board that reaches above average lifts the average the next board measures against. Economists borrowed a name for it from Garrison Keillor’s town where all the children are above average: the Lake Wobegon effect. Each board votes for something reasonable. The aggregate climbs anyway.

The peer group is where the quiet discretion lives. Reputable consultants screen for comparability, usually by revenue band, so the choice is not unbounded. But a board inclined to be generous decides where that band is drawn and which organizations sit at its edges, and can lean toward larger, better-paying systems without ever saying so out loud. The percentile target gets all the attention. The composition of the group beneath it does much of the work, and it does that work before anyone in the room votes on anything.

There is a reason nonprofit health systems lean on this method rather than more independent judgment. Federal tax law rewards them for it. Under the intermediate-sanctions rules, a nonprofit board can establish a rebuttable presumption that its executive pay is reasonable if the arrangement is approved by board members without a conflict of interest, the board relies on data about comparable organizations, and it documents that it did so. Set those three conditions beside the room and the safeguard starts to look thin. The conflict-free body is that same handful of finance-background members, and the comparability data is the benchmarking exercise itself. The rule written to keep nonprofit pay reasonable functions, in practice, as an instruction to benchmark. The very safeguard meant to discipline the number institutionalizes the mechanism that ratchets it. I would challenge us to sit with that inversion for a moment, because it is the clearest case I know of a structural feature producing the opposite of what it was built to do.

Congress has noticed more than once. In 2017 it placed a twenty-one percent excise tax on nonprofit pay above a million dollars, a measure that reads as a check on excess. A tax on the level, though, leaves the architecture that sets it untouched, the whole problem restated as tax policy.

The second feature is what the number lets the rest of us see, and what it does not. For a nonprofit system, the figure that eventually reaches the public sits on a Form 990, a tax filing few people outside the field ever open, and it appears only for a handful of the highest-paid people in the organization. Even that figure can understate the total, because incentive plans, deferred compensation, and other legitimate vehicles move real value into years and forms where it is harder to read in any single snapshot. There is a structural version of the same effect. A large system is rarely one organization but a family of them: some nonprofit and filing a 990, others taxable or for-profit affiliates that file no public return. Pay can be spread across that family in ways that are hard to add up, and for the parts that operate for profit there is no 990 at all. None of this is hidden, exactly. The 990s that exist are public and searchable, and the studies I am about to cite were built from them. It is disclosed the way a thing is disclosed when the disclosure is technically complete and practically unread. The opacity is not a conspiracy. It is the byproduct of an arrangement that was never built to make the community it serves a reader of the number.

Set the mechanism aside and look at the result. At twenty-two major nonprofit medical centers, the pay gap between hospital chief executives and registered nurses widened from twenty-three to one in 2005 to forty-four to one in 2015, while inflation-adjusted nurse pay barely moved (Marcus and colleagues, Clinical Orthopaedics and Related Research, 2018). A 2024 study in PLOS One found that most of the growth in nonprofit chief-executive pay tracks the size of the organization a leader runs more closely than the quality of the care it delivers, and it tied that growth directly to hospital consolidation. That second finding matters, because it means the incentive rewards getting bigger more reliably than getting better, and getting bigger is what the last decade of mergers and acquisitions delivered. A state-level analysis put it in human terms: a North Carolina report found the 2019 pay of eleven nonprofit hospital chief executives equal to the combined salaries of 572 registered nurses. Three numbers, and the last one is the one I cannot unsee.

Here is the objection I take most seriously, because I have watched the work up close and know it is real. Running a multibillion-dollar health system is genuinely hard. The people who do it answer for outcomes across dozens of facilities, they carry a financial weight that can sink an organization and everyone who depends on it, and they are recruited from a small pool by boards competing for the same few names. The leaders I am describing seldom touch a bedside, but they set many of the conditions every clinical team works under, and setting those conditions well is skilled, demanding, consequential work. So is the work on the floor. The nurse who catches a deteriorating patient two hours before the numbers would have, the hospitalist holding twenty admissions in her head at once, the tech who notices the thing on the monitor the protocol missed, each carries a weight measured one life at a time, and it is every bit as real as the enterprise kind. I am not arguing these leaders deserve little, or that the seat should be paid like the bedside. The market for that talent is real, and pretending otherwise would be its own kind of dishonesty.

That objection, though, defends the level of the pay. It does not defend the architecture. A person can be worth a great deal and the process that sets their number can still be a ratchet the community never voted for and rarely reads. Those are two different claims, and the room tends to answer the second by relitigating the first. The difficulty of the job is true. It is also, most of the time, the change of subject.

There is a tell in that phrase, voted for. A public company answers to shareholders who now hold an advisory say-on-pay vote, a chance to register that a number has run ahead of performance. A nonprofit health system has no such vote. The community underwrites its tax exemption, absorbs the consequences of its budget, and staffs its floors, and it holds no seat and no ballot on the number at all. That absence is not an oversight. It is the design.

I have spent the last few issues on a related question, who owes for the cost of a hospital staying ready to take the hard case at two in the morning. This room is the same problem from the other direction. The readiness a system underfunds and the number it protects are set by the same architecture, one that decides behind a closed door what the institution will carry and leaves the rest to the people who were never in the room.

So I keep coming back to the base number. It will shape a Tuesday-night floor eighteen months from now, and it is settled before the committee meets. The ratification, when it comes, can take less time than working up a single complicated admission. To test any of this from outside, the place to look is narrow and public: the peer group a system chose, which surfaces on its 990, and whether those organizations are truly its peers or a flattering reach upward. The board that sets the number rarely sees the three-in-the-morning version of the hospital it governs, and the hospital at three in the morning rarely learns what the number was or how it was set. Both are features of the same arrangement, and the arrangement is working as designed. That is not a reason for outrage. It is a reason to describe the room as it is, so the people who live with the number can at least see how it was built.

This is the fifth of the seven rooms I have been mapping; the map that holds them all lives here.

From the room you can’t see, the voice you need to hear.

Last week I argued that the number a hospital and its anesthesia group fight over every year, the subsidy, is not the real problem but the visible proxy for one. The real thing we are paying for is readiness, the standing capacity to handle the case that does not fit the outpatient model, and we pay for it now through a subsidy negotiated in the dark rather than a readiness payment named in the open. The most useful replies to that piece pressed on the question I had put off on purpose. Fine, they said: if readiness is the thing to pay for, who owes the payment? That is a fair question, and it is harder than it first appears. This is my attempt at an honest answer.

Start with why the question resists a clean answer. Readiness is close to a public good, and it behaves like insurance: the thing of value is the availability itself, the option the whole region holds on the chance that one of them needs the hospital that kept its doors staffed. The emergency department that stays lit before anyone knows who will need it, the team that can take the two a.m. complication, the obstetrics unit and the trauma bay and the intensive-care beds holding the patients too sick to be anywhere else, all of it is capacity the whole region draws on and only part of the region pays for. The surgery center a few miles away depends on that standby by law and pays for none of it: even now, a center has to keep a written procedure for transferring its emergencies to a local hospital, which means the low-cost site’s backstop on its worst day is formally the hospital next door, at no cost to the center. The healthy patient who went to the surgery center still benefits from that standby the day a bad outcome sends them back. Their plan benefits from it without buying it on the ordinary Tuesday. A community that has never once used its trauma bay is safer for its being there. When a benefit reaches everyone and the bill reaches only some, arguments about who owes get hard, because everyone at the table has a true reason to say it belongs to someone else.

There are four defensible answers to who owes, and each makes a real case. The commercial payer that steered the profitable case away has captured the savings and left the standby behind, so the party that captured the savings is the intuitive place to send the bill. The whole payer pool has a claim on it too, because Medicare, Medicaid, the self-funded employer, and the uninsured patient all lean on the same capacity, and a benefit that broad points toward a broad payment rather than a single check. The state has a claim, because standby is public infrastructure and a state like Maryland already sets hospital rates and funds capacity directly. And the community itself has a claim, because it is the community that needs the capacity to exist and, in the end, the community that loses when it lapses. Each of those is defensible. That is exactly what makes the question uncomfortable.

The answer that satisfies first is the commercial payer that moved the case. It is the party that made the choice and captured the gain, and the logic that the party who benefited should pay is neither radical nor new. The site-of-service difference that side of the ledger keeps is even larger in the commercial market than it is under Medicare, which is one reason a growing number of states have started writing site-neutral and facility-fee rules of their own. If you benefited when the profitable case left, the argument goes, you owe something toward the standby it left behind.

I find that answer attractive, and I think it is incomplete, for three reasons worth stating clearly. The first is double payment. The commercial payer already pays hospital rates that fold standby cost into every case it does buy inside the building, so a separate readiness charge risks paying twice for the same capacity. The second is the free-rider problem. If Medicare, Medicaid, self-funded employers, and uninsured patients all draw on the standby, singling out the one commercial insurer that happened to steer a given case is arbitrary, and a benefit that reaches everyone is the standard argument for broad public financing rather than a single-payer bill. The third is the perverse incentive. Charging a payer for readiness wherever it moves a case taxes the very migration most of us agree is right for the healthy patient, and it would prop up duplicated capacity that an honest accounting might say should consolidate. An answer that punishes the good outcome to fund the necessary one is not the answer.

The useful part is that we have already built a version of each of these answers, which means the real work is choosing among working models rather than inventing one. The narrowest and clearest sits where the pressure hit first. When rural hospitals began closing because volume alone could not sustain them, Congress created the Rural Emergency Hospital, which gives up its inpatient beds in exchange for a fixed monthly facility payment, roughly two hundred ninety-five thousand dollars in 2026 and adjusted each year, paid for keeping emergency and outpatient capacity available on top of what it bills for the care it delivers. That is the community-and-taxpayer answer, made narrow and explicit: a public payment for being ready, named as such. Maryland runs the broader version. For years the state has paid its hospitals a fixed global budget for a population rather than a fee for each case, which funds capacity instead of throughput, and at the start of this year that all-payer approach carried into a federal model that several more states are now adopting. That is the all-payer answer: everyone who draws on the system pays into the standby through the rates they already pay.

We even price standby in small, explicit pieces inside our own walls. The on-call stipend and the trauma-activation fee are readiness payments named for what they buy, and the fixed cost of trauma readiness alone, the coverage a center has to staff before the first patient ever arrives, runs into the millions of dollars a year at a busy center before anyone is billed for a single injury. The clearest evidence that the gap is real is the one place the law mandates readiness and declines to fund it: the federal requirement that hospitals keep an on-call panel available for emergencies arrives with no payment attached, and hospitals bridge it with stipends that rarely cover the cost.

Other industries are less shy about pricing the same thing. Electricity markets pay generators a separate capacity price for the promise to be available on the worst day, apart from the power they sell on an ordinary one, and in its most recent auction that promise cleared at a record price across the largest grid in the country as demand climbed. The principle is old and unremarkable everywhere except the hospital: standby is worth paying for on its own, and the parties that depend on it are the parties that owe.

I want to be careful not to suggest a tidy answer, because I distrust them, and because the record is candid about how hard this is. Pennsylvania put eighteen rural hospitals on all-payer global budgets for exactly this reason, to pay for the capacity a community needed rather than the volume it could no longer generate, and the results so far have been mixed, with margins that did not clearly improve and avoidable use that did not clearly fall. That is not an argument against naming who owes. It is a reminder that naming the payer is the first move rather than the last, and that a readiness payment built carelessly can be captured and padded like any other line by the same executives I have watched profit from a patient’s worst day. The case for building it in daylight is exactly that risk.

If I have to force an answer, and I think the question is worth forcing, it is this. Readiness is a public good, and public goods are paid for by everyone who draws on them, not by the single party that happened to move one case. That points toward the all-payer shape Maryland uses or the public facility payment the Rural Emergency Hospital uses, both of which spread the cost across the people who actually depend on the capacity. The satisfying answer, that the commercial payer who captured the savings should write the check, is the honest second-best, the cleanest stand-in when no all-payer mechanism exists, and it should be used with its flaws admitted rather than hidden. What we should stop doing is the thing we do now, which is to leave the bill on the hospital by default and call it a subsidy. The hospital does not simply swallow that bill, either; it recovers part of it by charging the commercial payers more on everything it still does, so a readiness payment is already being made, opaquely, folded into next year’s negotiated rates instead of named on its own line. We pay for it in the dark because the hospital is the one party in the arrangement that cannot walk away from the promise.

I lead an anesthesia company, so let me be clear, because a careless reading turns this into an argument for my own line item. It is not. My group is inside the same loss the hospital is, and the readiness I am pointing at is the hospital’s whole standby function, most of which sits far from my clinicians. Pricing it honestly, and naming who owes it, would end the annual ritual, repeated now at more than eight in ten hospitals that write some version of that check, in which a group and a hospital argue over a number that was never really about either of them, and would replace a private fight with a public accounting. That is worse for the negotiation, meaning worse for my group’s leverage inside it, and better for the institution, which is the trade I would make.

Here is the part that belongs to anyone who has done the work, whether that is the trauma bay at two in the morning, the clinic that holds the complex patient other settings turn away, or the floor team absorbing a surge of admissions on an ordinary Tuesday afternoon. Standing ready is the whole of that contribution, and it stands on its own without having to be justified upward; the institution is only poorer when it fails to listen to the people who carry it. The finance team can watch the subsidy line climb for years and read it as a cost to be managed between two parties in a room. The clinician who has carried the readiness knows what the line is actually buying, and can say what the ledger cannot: that it is not a courtesy and not a market distortion, but the price of a promise the community is counting on, currently billed to the wrong party because we lack the nerve to ask who really owes it. What each of us can do with that depends on where we sit. At the bedside, it is refusing the language of courtesy where the standby actually lives, and naming the readiness plainly in how the work gets described and defended. For anyone with a seat where the number is set, it is refusing to let the subsidy be argued as last year’s figure plus a percentage between a hospital and a group, and putting the real question on the table instead: who benefits from this capacity, who depends on it, and who therefore owes it, measured against the true cost of keeping the doors open. And it is carrying that question up to the level that can answer it, the system board, the payer at contract renewal, the state that sets the rate, rather than settling it in the annual room where a group and a hospital split a number that was never only theirs.

The healthy patient is still home before dinner, and that is still the right outcome; I would argue for it again tomorrow. The person who needs the hospital at two in the morning still needs someone ready, and someone still has to pay for that readiness whether or not we are honest about who. We can keep leaving the bill on the one party that cannot refuse it and calling that a subsidy, or we can do the harder thing and say the word out loud: who owes. Until we answer it, the standby stays an accident waiting on the night it finally fails, and the case that arrives that night will not care whose line item it was.

Subscribe now

In 2009 I was a couple of years out of residency, just beginning to learn the administrative side of an operating room, and my call nights had a rhythm to them. Somewhere after midnight the room would come to life for a perforated bowel: a patient who had come through the emergency department hours earlier, or arrived by transfer from a smaller hospital that had neither the surgeon nor the team to open an abdomen at that hour. The patient had planned none of it. They were sick and frightened and getting sicker, and between the decision to operate and the incision there was no time, and no process, for anyone to ask whether their insurance contracted with our facility. We operated. The coverage question waited for daylight.

Subscribe now

What I could not have told you then, standing at the head of that bed, is how the price of a night like that got decided. Neither the patient nor anyone else in the room had a hand in the number. And in that same season, though I was too deep in learning my own job to read it as a warning, the machinery that set such numbers was coming apart in public view in New York.

Seventeen years later, somewhere in the files of the federal arbitration system that now decides what out-of-network care is worth, there is a dispute in which an insurer reported, under a methodology federal rules currently permit, that the benchmark price for a high-acuity emergency visit was one cent. Behind a filing like that is a patient like mine: someone having one of the harder nights of their life, cared for by a team they never chose. And the filing has company: by a December analysis of the government’s own dispute files, more than sixty thousand disputes carried a reported benchmark under twenty dollars, and in over twenty thousand of those the number was zero. Whatever else those figures are, they are not the median price of anything.

A number like that reads as a clerical error until you learn its family history. I have needed the better part of two decades to appreciate how faithfully it repeats.

In February 2008, the attorney general of New York opened an investigation into how commercial insurers decided what out-of-network care was worth. At the center of it sat a database operated by Ingenix, a subsidiary of UnitedHealth Group, one of the country’s largest insurers. Most of the industry used that database to establish the “usual and customary” rate, the benchmark that determined how much of an out-of-network bill the insurer covered and how much landed on the patient. The investigation found what the ownership structure predicted: the rates ran low, through faulty data collection, poor pooling, and an absence of audits, and every dollar the benchmark understated was a dollar the insurer did not have to pay. The attorney general’s office was direct about the heart of it: a database presented to the public as independent was not independent at all.

The reckoning came in one January week in 2009. UnitedHealth first agreed with the attorney general to shut the Ingenix database down and put up $50 million toward an independent successor, FAIR Health, governed outside the industry it prices. Days later, it agreed to pay $350 million to settle the class action the database had spawned. The episode produced a lesson so obvious it keeps needing to be restated: whoever controls the benchmark controls the margin. A benchmark is not a neutral fact of nature. It is built by someone, from inputs someone chose, and when the builder profits from a lower number, the number finds its way lower.

Eleven years later, Congress had the chance to apply that lesson, and in one respect it did. The No Surprises Act, passed at the end of 2020, rightly protected patients from balance bills they never chose, and it needed a benchmark to make the rest of its machinery work. Lawmakers rejected the discredited “usual and customary” approach and chose something that sounds far harder to manipulate: the insurer’s own median contracted rate for the same service, in the same specialty, in the same market. That number, the qualifying payment amount, or QPA, now does two jobs. It generally sets what the patient owes in cost-sharing when out-of-network care arrives at an in-network facility, and it anchors the arbitration process, the one I wrote about in June, that decides what the clinician is ultimately paid. The diagnosis was right. The delegation is where it came apart. The statute handed the calculation of that median to the insurers themselves, with an opaque methodology and, as it turned out, very little checking of the math.

The construction rules do the quiet work. The median is anchored to rates in effect in January 2019 and carried forward by an inflation index, so rates negotiated in the seven years since do not reset the base. A plan needs only three contracted rates to form a valid median, which means a thin, carefully chosen panel can stand in for a market. Bonus and incentive dollars are stripped out before the median is taken. And the pool may include contracted rates for services the contracted clinician does not actually furnish. This last one, the ghost rate, does the heaviest lifting, and the arithmetic is worth seeing once in concrete form. If an insurer holds contracts with fifty primary care physicians whose panel agreements carry a fifteen-dollar placeholder line for anesthesia services, and contracts with ten anesthesia groups at negotiated rates around two hundred fifty dollars, the median across all sixty contracts lands on the placeholder. The fifteen dollars is a phantom: none of those fifty physicians administers anesthetics, and the line was not actually paid to anyone. It sets the benchmark anyway.

The size of the resulting gap comes from the payers’ own filings. As I wrote in June, a December study matched the QPAs insurers reported to the federal government against the median in-network rates the same insurers published in their own federally required transparency files, for the same service in the same market. The reported benchmark averaged one-third of the insurers’ own contracted rates, and sat below the insurer’s own published median in roughly two-thirds of disputes. The study’s authors are a provider-aligned coalition, and the transparency files are famously messy, so I hold the decimal points loosely. The direction is harder to dismiss, and the mechanism that was supposed to substitute for the subpoena this time has barely been used: since the system went live in 2022, the federal government has completed a single QPA audit.

Between the two eras sits a middle chapter suggesting the lesson stayed commercially alive while the vocabulary changed. Beginning around 2010, many payers routed out-of-network pricing through MultiPlan, a third party whose output carried the appearance of independence. Consolidated antitrust litigation now alleges the platform facilitated coordinated underpayment, nineteen billion dollars in 2020 alone by the plaintiffs’ estimate, with trial evidence they say shows payers could adjust the output. Those are allegations, disputed and unresolved, and I hold them at that weight. The pattern they describe, though, is the Ingenix pattern with better branding.

The courts have been circling the current version for four years. Twice the agencies tried to make the QPA the default answer, first as a presumption and then as a required starting point, and twice the courts struck the rule down, the Fifth Circuit writing that it placed a thumb on the scale in favor of the insurer-determined number. The live question, whether ghost rates and the incentive-payment exclusion are lawful at all, sits today before the full Fifth Circuit, which vacated a panel ruling that had upheld the methodology and has not yet issued its own. While it deliberates, the 2021 calculation rules remain in force under an enforcement grace period that runs through October 1 of this year. The benchmark anchoring a dispute volume that now runs into the millions is, at this moment, a number the courts have not finished deciding anyone was allowed to build that way.

The strongest objections deserve their full weight. Insurers point out that providers win the large majority of arbitrations, at multiples of the QPA, and read that as proof the process is being gamed. I’d argue the arithmetic points the other way: a benchmark that neutral arbitrators overrule most times they examine it is failing at its one job, and the disputes are only the visible edge, because the far larger volume of claims that do not reach arbitration is priced by the same number with no arbitrator watching. It is also true that some of the dispute volume comes from staffing firms, some private-equity backed and some in my own specialty’s neighborhood, that treated out-of-network billing as a strategy, and that conduct is part of what produced this law. The provider side of this fight does not arrive with clean hands.

I should be careful on two further counts. A three-judge federal panel did read the statute to permit the methodology, ghost rates included, before the full court set that opinion aside, so the legal question is genuinely contested rather than settled. And the empirical picture is less uniform than the headline numbers suggest: a Congressional Research Service analysis of emergency services found the median QPA above the median in-network rate in six states and below it in eight, with methods that differ enough from the December study to keep the debate honest. Based upon what we know at the moment, the fairest statement is that the benchmark runs low in most places it has been checked, and that it has barely been checked. The design critique survives even the most charitable reading, because it does not depend on any particular gap in any particular state. The party with a direct financial interest in a lower number holds the pen that writes it. In 2009 we called that arrangement a scandal. In the current version it is a compliance methodology.

I lead an anesthesia company, and anesthesia sits with emergency medicine and radiology among the specialties most exposed to this number, so I am not a neutral party to it, and I have tried to write this in a way that does not need me to be. This is not only a specialist’s fight: the same benchmark helps set what any patient owes on an out-of-network bill, which is why it belonged to the public long before it belonged to a docket. The comparison that matters is not between what providers want and what insurers pay. It is between what the insurer reports the benchmark to be and what the insurer’s own fee schedules say the market is. In 2009 it took an attorney general with subpoena power to put those two numbers side by side. Today they sit in public files, a download apart, for any clinician or executive willing to run the comparison for their own specialty and their own market. Seventeen years ago that comparison shut a database down, and the lesson was supposed to be that a benchmark cannot belong to the party it prices for. Sometime soon, the full Fifth Circuit will tell us whether that lesson holds as law, and whether the next penny in the files is an error or a price. Tonight, in some hospital, an operating room is coming to life for another patient, and the coverage question is again waiting for daylight. We have settled this before. The only question is whether we meant it.

Subscribe now

That case is also the migration in miniature. For most of my career, the smartest people in healthcare have been working to move care out of the hospital. We pushed procedures into ambulatory surgery centers, visits into outpatient offices, and recovery into the home, and for patients the results have largely been good. Care got cheaper, access got easier, and the setting around a routine procedure grew simpler and less frightening. I have argued for that migration myself, in strategy rooms where the math was plain and the benefit to patients was real.

What we discussed far less, because it was harder to see, is what that migration does to the building we are emptying.

A hospital loses money on most of what a community needs it to do. The emergency department, the behavioral health unit, charity and uncompensated care, obstetrics in a rural county, the residency program that trains the next generation: few of these cover their own costs, and several were never built to. For a generation, the operating room has quietly paid for them. Surgical and procedural services have been the hospital’s engine, generating the margin that funds nearly everything the institution loses money on, and most of the people who work on the surgical floor have little reason to watch the gauges. By one industry estimate, surgical and procedural services generate roughly sixty percent of a hospital’s operating margin. That is the engine I mean.

The engine is losing power at the same moment the organization is leaning on it harder. That is the whole of the problem, and it is worth saying plainly before anyone mistakes the symptom for the disease.

It is fair to note, before the finance reader does, that surgery is not the only quiet engine. In many non-profit systems the 340B drug-pricing program has grown into an offset of real consequence, and that room deserves its own walk-through another week. But the lever is available only to some hospitals, and it is under mounting pressure of its own, so leaning on it buys time rather than a trajectory.

Let me be clear about where I stand, because it shapes how you should read the rest of this. I lead an anesthesia company, and my group is not standing outside this problem pointing at it. We are inside it, losing to the same force the hospital is losing to. When the simpler cases leave for the surgery center, the hospital loses the margin that paid for its emergency department, and my clinicians are left with the harder cases, the longer nights, and a workforce that was already too thin. The subsidy a hospital pays an anesthesia group is the number everyone fixes on, and it is the symptom rather than the disease. I am not writing to argue that anyone should pay anesthesia more. I am writing because the case leaving the building is the event that matters.

Four forces are compressing the surgical margin at once, and each of them is a reasonable response to how someone is paid.

The first is the migration itself. High-margin elective surgery keeps moving to freestanding ambulatory centers because commercial payers prefer the lower-cost setting and because refined laparoscopic and robotic technique keeps expanding what can safely be done there, and Medicare’s push toward site-neutral payment, alongside the steady expansion of what it will pay for in an ambulatory center, is eroding the premium hospitals once earned for doing the same procedure under their own roof. The hospital’s advantage is narrowing by design.

The second is what stays behind. As the straightforward, well-insured cases follow the market out the door, the hospital operating room keeps an older, sicker, more Medicare-weighted population. Researchers studying hospitals that lost volume to ambulatory centers found a five percent rise in case-mix index and a ten percent rise in average length of stay between 2019 and 2023. The cases that remain cost more to perform and pay the institution less.

The third is the cost of the people who make surgery possible. More than eighty-five percent of hospitals now write a check to subsidize their anesthesia groups, and even ambulatory centers have begun to: the share expecting to pay an anesthesia stipend climbed from twenty-eight percent in 2024 to forty-four percent in 2025. This is a story about arithmetic rather than appetite. Medicare’s anesthesia payment fell from $22.27 a unit in 2019 to $20.43 in 2024 while clinician compensation rose by double digits, and national benchmarks now put median anesthesia compensation per unit slightly above median collections per unit. The subsidy exists because the work, paid at today’s rates, has stopped covering its own cost. That is the symptom in its clearest form.

The fourth is demand for that same scarce workforce in more places at once. Anesthesia is increasingly required outside the operating room, in cardiac catheterization labs, endoscopy suites, and interventional radiology, where structural heart procedures and advanced endoscopy have multiplied. By claims-based analyses, anesthesia delivered outside the operating room rose from about twenty-eight percent of all cases in 2010 to forty-three percent in 2023, with more than half projected within the decade. The same shortage that strains operating-room coverage now strains coverage everywhere else the hospital hopes to grow.

Put those forces together and you get a financial story that finance alone fails to capture. The part the margin tables miss is what the migration does to the hospital as a place to work.

When the easy cases leave, they take with them the rhythm that made the schedule survivable. The quick, predictable, daytime procedure that let a team breathe between hard ones is now done across town at a center that closes at five and carries no call. What remains inside the hospital is the heavy, complex, after-hours work, concentrated. The surgery center runs no emergency board at two in the morning, and the outpatient office is dark on the weekend, so the nights and the genuinely sick belong, more than ever, to the people who staff the hospital. The same pressure reaches well past anesthesia: the hospitalist, the intensivist, the palliative physician, and the bedside nurse inherit that concentrated acuity too, tending an inpatient census that has grown sicker on average because the well and the straightforward are now cared for somewhere else. For hospital-based anesthesia, the call burden and the unsocial hours have always been the hardest part of the job, and the migration deepens precisely that burden while a retiring workforce thins the bench that carries it. Roughly fifty-seven percent of practicing anesthesiologists are over fifty-five, and the training pipeline is capped by design rather than by interest: in the 2025 match, more than three thousand applicants competed for fewer than nineteen hundred anesthesiology residency positions. The hospital is asking a shrinking group to absorb a harder schedule, and that is a recruiting and retention problem long before it becomes a budget line.

It would be easier to write this if there were a villain, but there is not one. The commercial payer steering a knee replacement to an ambulatory center is lowering the cost of care, which is what we asked it to do. The health system cannot block the migration without forfeiting the cases entirely. The ambulatory operator captures the margin without inheriting the obligation to keep an emergency department open at three in the morning, which is architecture rather than character. Everyone is behaving rationally inside the incentives they were handed, and the sum of all that rational behavior is a hospital slowly losing the means to fund the services a community cannot do without.

This is why the fight over the subsidy is the wrong fight, or at least the small one. The subsidy is the thermometer. It rises because the case that used to run a margin now runs a loss, and arguing over the number treats the measurement rather than the fever. The disease is the departure of the case and the hollowing of the room it leaves behind. A hospital that negotiates its anesthesia subsidy down by a few points and calls the problem solved has cooled the thermometer and left the trajectory untouched. The question worth the energy in the room is the larger one: what replaces what the surgical margin has been quietly funding, and how does the institution keep its inpatient work survivable enough that people will still choose to do it?

I do not have a tidy answer, and I would be wary of anyone who offered one, because the forces involved are structural and they are converging by design rather than by accident. What I would ask of the leaders who sit where I sit is narrower, and I think harder. See the trajectory early, while there is still room to act, rather than waiting for the subsidy line to force the conversation. Acting early looks unglamorous and specific: re-engineering the inpatient schedule so the complex work that remains is survivable to staff, building coverage for the procedures that have moved outside the operating room deliberately rather than reactively, and pricing the subsidy conversation against the true cost of keeping the doors open rather than against last year’s number. Name it honestly to the board, including the part that implicates our own comfortable assumptions about growth. And stop spending the institution’s scarce attention on the thermometer when the fever is the thing that will take the building down.

We were right to move care closer to the patient. The migration made care cheaper and kinder, and I would argue for it again tomorrow. We were wrong to assume the engine room could keep funding everything else while we emptied it of the work that kept it running. The hospital that comes through this decade will be the one whose leaders saw, early enough to matter, that the case leaving the building was never only a billing event. The finance teams could watch the margin fall for years. What the people in the room and at the bedside could see, and say first, was what it was doing to the work itself.

Subscribe now

He chose none of us. He did not pick the emergency physician who evaluated him, the radiologist who read his scan, the anesthesiologist who carried him through the operation, or the pathologist who confirmed his diagnosis. In most of medicine the choosing is half the relationship: you select your surgeon, you read about the practice, you ask a friend who delivered her babies. The specialties that gather around the sickest and most sudden moments are the exception, and that single fact sits underneath one of the quieter fights in American healthcare.

The No Surprises Act got something important right. It decided that a patient who never chose the out-of-network clinician should not be handed the bill for that accident of scheduling, and that was the correct call. In the same motion, the law took away those clinicians’ ability to bill the patient for the balance and replaced it with a single federal arbitration process, independent dispute resolution, as the only road left to argue that the insurer’s payment was too low. So the surprise bill the patient used to receive became a dispute the clinician now files against the plan. The patient is protected. The argument over what the care was worth did not disappear; it moved to a different room.

Most coverage reads that process as a billing squabble, a story about paperwork and arbitrators and fees. I’d argue it is something larger and simpler: a proxy war over who captures the margin on that asleep patient’s care. And the way to tell who is winning is to stop watching the arbitration and watch the benchmark, the single number the rules let one side calculate.

Earlier this month, on June 4, the Departments of Health and Human Services, Labor, and Treasury finalized the long-awaited rule governing how the dispute process works, and the improvements are real. The administrative fee to bring a dispute fell from $115 to $15. Providers can now batch as many as fifty related claims into a single case. Insurers will have to use standard codes to signal, up front, whether a claim is even eligible for the process. The societies that speak for anesthesiology, emergency medicine, and radiology welcomed the rule, and they were right to. When a single radiology claim is worth a modest sum, a $115 fee made it irrational to challenge a payment everyone knew was too low, because the dispute cost more than it could recover. At $15, the arithmetic changes, and a small practice can afford to show up. That is a genuine step toward a process a community group can actually use, and it deserves to be called what it is.

But a process can be cheaper to enter and still rest on a foundation that decides the outcome before anyone walks in. The new rule does not touch the qualifying payment amount, the QPA, the benchmark the arbitrator leans on when deciding what a service was worth. And the benchmark is the whole game. An analysis from Americans for Fair Health Care, a provider-aligned coalition, compared the reported QPAs against insurers’ own published median in-network rates, so the comparison rests on the insurers’ own numbers rather than the coalition’s. In roughly two-thirds of disputes the reported QPA came in below the insurer’s own median contracted rate for the same service in the same region, averaging about a third of the real rate. A median that sits below the real median sounds like a contradiction until you see how the number is built. Tens of thousands of disputes carried a QPA under twenty dollars. Some carried a QPA of zero. Hold that next to the figure insurers cite most often, that providers win the large majority of disputes, and the win rate stops sounding like abuse and starts sounding like arithmetic. When the benchmark is set at a third of the market, an award at three times the benchmark is a correction, not a windfall.

The mechanism that pushes the benchmark down has a name a careful reader will want: ghost rates, the practice of folding into the median the contracted rates for services that were never actually performed, alongside rates borrowed from adjacent specialties that never did the work. It works the way a salary survey would if you padded it with people who were never paid for jobs they never held: drop in enough zeros and near-zeros, and the median sinks toward them. Whether that practice is lawful is the question now before the full Fifth Circuit. A three-judge panel first let the methodology stand; the full court then agreed to rehear the case, which set that panel decision aside, and it has not yet issued its own ruling. While the court takes its time, the Departments have allowed the existing benchmark methodology to remain in force through this fall. So the plumbing of the dispute got cheaper and clearer in June, and the number the dispute turns on stayed exactly where it was, contested and unresolved. That is not an accident of timing. It is the part of the fight that actually moves the money, and it is the part still unsettled.

Follow the money one step further and the proxy war stops reading as a metaphor. A Health Affairs analysis last fall described a mechanism that, if it operates at any scale, ought to change how we read the entire dispute: many plans, or the third-party vendors they hire, take a percentage of the spread between a provider’s billed charge and what the plan ultimately pays. Billed charges are set by the provider rather than by any market, so the number itself is soft, but the incentive it creates is not: the wider that spread, the larger the plan’s own cut. An out-of-network claim produces the widest spread there is, so the program pays the plan more to push anesthesia, radiology, and emergency medicine out of its network than to keep them in it. Put plainly, the more a plan pays a clinician, the smaller its own internal bonus. Seen that way, network contraction is not always the unfortunate byproduct of hard bargaining. In some cases it is the design, and the dispute that follows looks less like a breakdown than like the system working as intended.

I want to be careful here, because this is the point where the argument is easiest to overstate. The mechanism is documented; the scale is not. We do not yet have clean, specialty-level data on how many out-of-network terminations were deliberate rather than incidental, and an honest case names that gap instead of papering over it. Based on what we do know, though, the direction of the incentive is not in question, and it turns the insurer’s central complaint on its head. A party that is paid to widen the very gap these disputes exist to close is not a bystander to the volume. It is closer to the source of it.

The honest objection, the one a well-prepared opponent leads with and the one I’d challenge us to sit with rather than wave off, is that a small number of private-equity-backed arbitration firms file an outsized share of these disputes. Three filers account for something close to half the volume, and one of them has won at award levels that, on their face, look like a machine built to extract. Insurers point at those firms and ask Congress to treat the entire process as abuse. The conflation is their strongest move, and it deserves a real answer. Two things are true at the same time. A firm winning at nine times the QPA is, before it is anything else, evidence that the QPA is broken, because when the benchmark is a third of market, large multiples are the expected result rather than proof of gaming. And the concentration itself exposes an access problem the new rule only begins to address: the parties who can work this system at scale are the ones who can fund a standing dispute operation. The independent anesthesiologist in a community hospital, the clinician the law was nominally written to protect, was until this month’s fee cut largely priced out of the process built in her name. The point is not that every high-volume filer is virtuous. It is that the volume problem, the benchmark problem, and the network problem all share one author.

There is one argument I will not set aside. Self-insured employers, not insurer shareholders, ultimately pay the awards that come out of this process, and a system that moves large sums from employer health plans to aggregators raises a real distributional question that the corruption of the benchmark does not answer. Both can be wrong at once. A piece written honestly from the provider side has to hold that, and I do.

Which brings me back to the man who came through the emergency department on a Tuesday, asleep before he could have weighed in on any of it. The new rule lowered the cost of entering the fight on his behalf, and that matters. It did not make the fight fair, because the number that decides it is still calculated by one of the two parties to it, under a methodology the courts have not finished examining. A cheaper seat at a tilted table is progress. It is not the same thing as a level table, and naming that difference out loud is work any clinician close to it can do, and work the people in leadership exist to carry forward. In practice, that begins with a small, concrete ask: to see how the benchmark was built before the argument turns to what any one award should be. It is the hospitalist who can no longer find an in-network specialist to accept the consult, and the primary care physician left holding what the narrowed network pushes back onto her, who feel these decisions in their own work long before they reach a policy paper. The question worth answering is not whether providers win too often. It is who was allowed to set the number, and why the people closest to the patient had the smallest hand in it.

I have done both versions. I have also watched, in colleagues I admire and in myself, the morning when the ten-minute version quietly and insidiously disappears and only the four-minute one is left. The questions still get asked. The evaluation still gets done. The hands are as good as they ever were. What goes missing is the part where the patient stops being the next case, and the unsettling thing about its absence is how reasonable it feels from the inside. You are tired in a way that sleep does not fix. You have been asked, too many times, to move faster than the work deserves. Faster than the patient in front of you deserves. So the caring is the thing that burns off first, because it is the most expensive thing to keep spending.

That loss has a clinical name, and naming it correctly is the whole point of this piece. Christina Maslach’s inventory, the instrument most of medicine uses to measure burnout, breaks it into three parts: emotional exhaustion, a diminished sense of accomplishment, and depersonalization. The first two describe how the clinician feels: tired and insignificant. Depersonalization is different, because it describes what happens to the people on the other side of the clinician. It is the cynical, impersonal distance that clinical training spends years trying to prevent, and it is the part of burnout that lands most directly at the bedside. You can see it anywhere clinicians work: the clinic visit that ends a question early, the hospital room visit that no longer pauses, the hard consult delivered without the silence the news deserves. An exhausted clinician may still be trying. A depersonalized one has quietly stopped expecting the encounter to matter, and the patient can feel that even when they cannot label it.

The distance is not only a loss of warmth. It travels with measurable harm. A meta-analysis pooling thirteen studies and more than twenty thousand physicians found burnout associated with roughly a threefold increase in the odds of a self-reported major medical error, with depersonalization contributing independently rather than only riding along on exhaustion. And the burden falls across the whole team, not one credential at one end of the hallway. Burnout among certified registered nurse anesthetists, nurses, and advanced practice providers all runs at or above the physician rate in recent national surveys; the instruments and the survey years differ, so those numbers are best read as separate readings rather than a clean ranking, but the direction holds across every credential on the team. My own specialty felt a particular version of it: the share of anesthesiologists reporting workplace staffing shortages more than doubled in two years, from about a third in 2020 to nearly four in five by 2022.

Here is the finding that should change how a leader thinks about all of it. Depersonalization is contagious. A study of intensive-care nurses found that the burnout of the people around you, and specifically their detachment, predicted your own better than the organizational stressors we usually blame. It is a single study, and an association is not proof of transmission, but the finding cuts against the frequent and comfortable assumption that culture flows only from the top. Cynicism turns out to be as catching as exhaustion and considerably harder to see. One worn-down clinician who has stopped expecting the work to matter does not stay one clinician. The posture moves down a unit the way a mood moves through a house, and a culture forms around it: clock in, get through the list, sign out, protect yourself. This is the part the wellness brochure cannot reach, because it was an organizational problem from the start.

The name that has attached to that culture is the shift-worker mentality, and it deserves to be handled with some care, because the phrase carries a sneer it has not earned. It entered medicine two decades ago as a worry about resident duty-hour limits, the concern that capping the hours would produce trainees who watched the clock and felt little ownership of their patients. I am borrowing the term for something those original papers did not quite describe: the experienced clinician, years past training, who has narrowed the job down to its hours on purpose. Said that way, it sounds like a character flaw. It is closer to arithmetic. When the conditions of the work repeatedly ask you to invest more of yourself than the conditions will protect, reducing the investment is a rational decision and not a moral one. It is a very special form of self-protection.

And the conditions that wear a person down are not only the large clinical compromises we name most often. They are the small subtractions of control that pile up underneath them: the clinic or OR start time you cannot push back an hour to take a child to school, the shift you cannot trade, the call you cannot give away. Your own schedule has become a thing you request rather than arrange. Lose enough authority over the shape of your own day, down to the hour, and you learn which parts of yourself the institution means to let you keep, and you stop offering the rest. The clinician who clocks out clean, who has decided the institution will get competence and skill but not the last unguarded part of them, has usually reached that decision for good reasons, and treating it as a failing is both wrong and a guarantee that it deepens.

Which brings up the reframing I find most useful, and the one most likely to make an executive uncomfortable, because it moves the responsibility. In 2018 two physicians, Simon Talbot and Wendy Dean, argued that we have been reaching for the wrong word. Clinicians, they wrote, are not mostly burning out, a phrase that locates the fault inside the worn-out individual. They are suffering moral injury: the particular damage of being required, repeatedly and by people with authority over them, to deliver care they know is worse than the patient deserves. Burnout asks what is wrong with the clinician. Moral injury asks what is being done to them. The distinction is not academic, because it decides where you aim the remedy. If the problem is burnout, you send the clinician to a resilience workshop. If the problem is moral injury, that workshop reads as an insult, because it treats the symptom and leaves the wound exactly where it was.

I want to be fair to the other side of this. Personal resilience is real, the clinicians who guard their sleep and their relationships and their few quiet hours genuinely fare better, and a leader who tells people none of that matters is lying to them. The moral-injury frame is not a permission slip that absolves each of us of any responsibility for our own well-being. But as the primary institutional answer, wellness programming has been studied enough now to be honest about it: the reviews consistently find that organizational changes outperform individual ones, and that resilience training offered in place of fixing the conditions is, at best, a comfortable way to be seen doing something. Talbot put it more precisely than I can, suggesting that burnout is often the end stage of a moral injury that went unaddressed.

So how much can the person in the executive seat actually change, the clinician who crossed into administration and now signs the productivity targets, set against the colleagues I respect who weighed the same move and chose to stay at the bedside on purpose and keep their hands on the work? More than the executive usually admits, though the reach is more indirect than the title implies. Tait Shanafelt’s group at Mayo found that each one-point gain in a supervisor’s leadership score was associated with a three percent drop in the odds of burnout among the people who reported to them, and that at the work-group level, half of the variation in satisfaction traced back to the behavior of the immediate leader. Half. That finding is about the front-line leader one rung up, the charge nurse or the medical director, not the corner office. Which means the executive’s real lever is rarely direct: it is whether those front-line leaders are chosen, trained, and shielded well enough to lead that way. And a one-point gain is not charisma. It is the leader who holds a real debrief after a hard case instead of moving straight to the next one, who carries an unreasonable target upward rather than passing it down, who still learns the names.

That finding cuts both ways, which is the part worth sitting with. The physician leader who enforces a number she knows is unreasonable, who reframes a clinical objection as an operational complaint, who models the very detachment we have been describing, is not a bystander to the spread. She is a carrier. The same seat, used differently, to take real administrative weight off the people doing the work, to say plainly and clearly that what they are carrying is injury and not weakness, to absorb a budget miss rather than pass the pressure down to the bedside, is one of the few real buffers the system has. The leader in that seat is either a carrier of the detachment or a buffer against it, and the people on the team can always tell which.

This is where the personal becomes a community problem, and that turn is the part the wellness conversation leaves out entirely. A clinician who has gone quiet on the inside often leaves for real before long, and the leaving is not evenly distributed. More than a hundred thousand nurses left the workforce across two years, and something close to four in ten of those who remain say they intend to within five. In a city, a single departure is absorbed. In a rural county, where most of the map is already medically underserved and the physicians are older than the ones available to replace them, one surgeon’s resignation can end surgical access for an entire region. The detachment that began as a tired clinician’s rational self-protection, the fault of the conditions and not the clinician, ends several steps later as a town that has to drive two hours to deliver a baby. That is the community cost, and it is the reason this is not a soft subject.

I keep coming back to the four-minute interview and the ten-minute one. The difference between them is the whole thing. It is where trust is built, where the catchable error gets caught, where a frightened person is steadied, where a young clinician learns that the encounter is the work rather than the obstacle to it. When the caring burns off, that difference is the first thing to go and close to the last thing anyone thinks to measure. We have spent years asking clinicians to be more resilient about losing it. The more honest question, and the one that belongs to anyone with the authority to change the conditions, wherever they sit, is what we are doing to the ones that made the ten-minute version feel like a luxury in the first place.

The economics of how those conditions came to be is its own subject, and the next thing I want to write about here. But the human version comes first, because the human version is what a community actually loses when its clinicians go quiet: not a line on a turnover report, but the unhurried, expensive, irreplaceable minute in which a clinician decides to let a stranger matter.

Past the Door publishes free pieces every Sunday. If this one named something you have felt, forwarding it to a colleague carrying the same weight is the best endorsement.

Subscribe now

I am David Wild. I have spent more than two decades moving between two kinds of rooms: the clinical ones, where I still practice anesthesiology a few shifts a month, and the administrative ones, where the conditions for that clinical work get set, the boardroom, the credentialing committee, the contracting table. Most clinicians only ever stand in one kind. Past the Door is my attempt to describe the other ones plainly, and to trace the line from each of them to the bedside, where the consequences land.

If you read one thing first, read the map.

Download the one-page field guide (PDF)

The seven rooms you cannot see is the anchor of everything here. It names the seven closed rooms where healthcare’s decisions actually get made, and gives each one a line. The image above is the keepable version, a single page you can save or send to a colleague. The full piece, with the sourcing, is the pillar it comes from.

Where to begin reading

The map orients; the room pieces go inside. A few places to start, depending on what brought you here.

• For the origin, start with the inaugural, The rooms you can’t see. It is the three-in-the-morning version of why I write this.

• For a room in depth, start with credentialing, the first one mapped fully. What You’re Actually Asking For is the applicant’s side; Inside the Credentialing Committee is the room itself.

• For the leadership thread, What Clinicians Get Wrong Moving to Administration, written with Tracy Young, is on the translator role and the identity cost of the move, and The Coach Who Became a Mentor sits beside it.

• For why any of this matters to me, After the Transplant is the most personal thing here.

The rest of the rooms, contracting, the executive office, peer review, compensation, succession, and the deal room, each get their own piece in the weeks ahead. Subscribing is how you get them as they land.

One ask, and it is a real one. Reply to the welcome note, or leave a comment below, and tell me which room you most want to understand. I read every response, and it shapes what I write next.

Subscribe now

From the room you can’t see, the voice you need to hear.