The disappearance of the independent physician.

Physicians are losing ownership of their practices, and patients are losing a physician whose only interest is their care.

An institution that lasted for most of the history of American medicine, the physician who owned the practice and had no employer, is ending within a single working generation. In 2012, 60.1% of American physicians worked in a practice that physicians owned. By 2024, the share was 42.2%. Over the same period, the share of physicians who held any ownership stake in their practice decreased from 53.2% to 35.4%, and the largest decrease was among physicians under 45, who are now the least likely of any age group to own the practice in which they work. The decline will continue as the generation that still owns retires and is replaced by one that never owned at all.

Behind that decline is a cost that has grown out of all proportion to the care it supports. From 1975 to 2010, the number of physicians in the United States grew 150%, while the number of health care administrators grew 3,200%. The apparatus that grew is the one that exists to get a practice paid by insurers, and it has reached a size at which its cost now decides whether a practice can survive on its own.

I am a physician who left clinical medicine for the work of removing this cost from independent practices, because I had come to think it is what decides whether they stay independent.

The loss matters because ownership decides whose interests affect the physician's decisions. A physician's purpose is to do what is best for the patient, and no physician pursues it without limits, because insurers decide what they will pay for and require approval before much of the care is given. Those limits exist whether the practice is owned by physicians or by someone else. When physicians own the practice, the only interests present beyond the insurer's are the patient's care and the physicians' own income, and that income depends on caring for patients well. When the practice is owned by a hospital system, a corporation, or a private equity firm, the owner holds interests the patient does not share: a return on the money it paid for the practice, the number of patients seen in a day, the cost of the care, and where referrals are sent. The owner advances those interests through how it pays the physician and the rules it sets. A sale leaves the insurer's limits as they were and adds an owner whose financial interest now affects decisions that the patient's care alone used to determine.

The added interest changes the care patients receive. In a survey of 1,000 employed physicians, only 18% said that the growth of corporate ownership has improved the quality of care. Nearly half, 47%, said that their employer's policies or financial incentives led them to change a patient's treatment in order to reduce its cost, and 70% said that their employer rewards or penalizes them according to the number of patients they see in a day. Consolidation also raises what patients pay. After a hospital acquires a practice, the price of that physician's services is 14.1% higher, because a hospital that has bought the independent practices around it faces less competition and can charge more. The higher price produces no improvement in the quality of the care.

Physicians have made the case for ownership directly. Bradford W. Holland, the president of the Texas Medical Association, argues that the care delivered by physician-owned practices, organized around the patient, has been among the most efficient and the highest in quality that American medicine has produced. His association makes the same argument about hospitals: the physician-owned kind competed with the large systems and lowered prices until a provision of the Affordable Care Act barred new ones in 2010, and the ban removed that competition.

Physicians sell anyway, and the reasons they give are financial and consistent. The reason cited most often, by 70.8%, is the difficulty of negotiating adequate payment rates with insurers, which a larger organization does from a stronger position. The next, at 64.9%, is the need for expensive equipment and systems that a small practice cannot afford on its own. The third, at 63.6%, is the burden of meeting the regulatory and administrative requirements that insurers and the government impose. All three come from a single cause, the cost of the administrative labor that every patient visit now requires.

That cost has become harder to afford every year, because the revenue available to a practice has not increased in proportion to its expenses. The inflation-adjusted value of the payment Medicare makes to physicians is now 33% lower than it was in 2001, and private insurers commonly use Medicare's rates as the basis for their own. Across one five-year period the Medical Group Management Association examined, the operating cost of supporting a single physician increased 29.3% while the revenue that physician produced increased 16.6%, and close to 9 in 10 practices report that their expenses are still increasing.

The labor costs what it does because of what it consists of. Running a practice is a set of distinct processes, and each one is a sequence of steps performed by several different people. Before a patient is seen, one person schedules the visit, another confirms that the patient's insurance is active and will cover the care, another obtains the insurer's advance approval for any service that requires it, and another gathers the outside records, prior notes, and test results the physician will need and assembles them into the chart. During the visit, the physician examines the patient, decides on a course of treatment, and writes the note. After the visit, one person reads the note and assigns the billing codes, another submits the claim to the insurer, another works the claims the insurer refuses and submits them again, another sends and tracks the referrals the physician ordered, and another obtains approval for the tests and medications that require it. Alongside all of this, faxes and records arrive continuously and have to be read and routed to the right place, and quality and risk reports have to be compiled and filed. The processes are also connected to one another, because the result of one becomes the input to the next: the records gathered before the visit are what the note is written from, the note is what the codes are taken from, the codes are what the claim is built on, and a referral sent out of one practice becomes a record that arrives at another. No one person performs a whole process from start to finish, and no process stands on its own, so a practice has to employ people both to perform each step and to move the work from one step to the next.

The largest part of the cost is the staff devoted to getting paid, because collecting what an insurer owes for a visit has become a full occupation in itself. A practice now employs 5.85 support staff for every full-time physician, and a large share of them do no clinical work at all. The most time-consuming task, which insurers call prior authorization, is obtaining the insurer's approval before a treatment the physician has already judged necessary can be given. A practice submits an average of 39 such requests for each physician every week and spends about 13 hours of physician and staff time on them, and in 40% of practices someone is employed to do nothing else. The faxes belong to the same work: in 2026, the fax remains a principal means by which one office sends a patient's records to another. None of this work produced any payment a generation ago. It exists because an insurer pays nothing for a visit until the visit has been documented, coded, and justified, and because decades of accumulated regulation, together with the move to electronic records, have added reporting requirements that have no effect on the care a patient receives. Across the health care system as a whole, work of this kind now accounts for between a quarter and a third of every dollar the country spends on health care.

The cost of the work is one problem, and staffing it is another. Most of it is done by medical assistants, an occupation that is at once among the fastest growing in the country and among the hardest to keep filled. About 811,000 people work as medical assistants. The occupation is projected to grow 12% over the next decade, 4 times the 3% average for all occupations, and about 112,300 positions open every year, many of them because the people in the job leave it. The work is demanding and the median pay is $44,200 a year, or $21.25 an hour, so a practice can often afford only people who are still learning it after they are hired. The accuracy of a patient's record and the payment of a claim therefore depend on people in one of the lowest-paid and least stable positions the practice has. When one of them leaves, the practice has to find and train a replacement and, in the meantime, handle the errors and the unfinished work that accumulate.

After the practice has paid for all of this, little remains. The salaries and benefits of the support staff alone take close to a quarter of everything a practice collects, and once rent, supplies, and the cost of software and reporting systems are added, overhead amounts to nearly 60% of revenue. The margin left is about 8% in a typical practice, and it is decreasing, so an increase of even a few percentage points in cost is enough to eliminate a large part of a year's profit.

The work the practice cannot afford to staff is done by the physician, in the physician's own time. The average physician now works 57.8 hours a week, and 20.3 of those hours are taken by documentation, by orders and results, by referrals, and by insurance forms and requests for approval, and none of them are spent with a patient. The hours are only part of the cost. The work is clerical, the entry of the same facts into one form and system after another, and it occupies the physician's attention while requiring none of the training that took a decade to acquire. Every hour spent on it is an hour not spent with a patient, so it also limits the number of patients the practice can see and the income it earns from caring for them. When a physician can no longer reconcile the hours the work takes with the money the practice earns, two options remain: keep the practice independent and do the administrative work without help, or sell to an organization large enough to cover the overhead and to negotiate better rates with insurers. That physicians have so consistently chosen the second option is the reason ownership has decreased as far as it has.

All of the costs that force that choice are one cost. The staff devoted to getting paid, the turnover among the people who perform the work, the overhead that takes nearly 60% of revenue, and the 20.3 hours a week of the physician's own time are payments for the same administrative labor.

One of the reasons physicians give for selling appears to stand apart from this cost. The most common, cited by 70.8%, is the weakness of a small practice in negotiating rates with insurers, and that weakness is a question of size. A large organization is paid more for the same work because it negotiates from a stronger position, and no efficiency inside a small practice changes that.

Size and ownership can be held together. Physicians have long combined into larger entities to negotiate while keeping ownership of their own practices, in independent physician associations and in groups that bargain as one body for the practices beneath them. There is already a class of companies built to supply that scale to practices that stay independent, of which Privia Health is the most prominent, and they give a small practice the negotiating position of a large one without a sale.

These arrangements pool the administrative labor and spread its cost across many practices. The billing, the prior authorizations, and the records still pass through people, and the platform performs that work for many practices at once, so the cost per practice falls while the payroll remains. Sharing a cost across more practices lowers it; it does not end it. The cost ends only when the work is no longer performed by anyone, which is a different thing from performing it together at scale. The weakness of the small practice is real, and removing the labor that every practice still pays for, large or small, closes the gap at its source.

For a long time that expense looked like a fixed condition of practicing modern medicine. It is not. What I have learned from removing it is why a decade of software built to reduce the expense has not reduced it.

The expense is a payroll. Every dollar of it is the salary of a person hired to perform steps and to move work between steps, and a salary is paid in full or it is not paid at all. A practice cannot employ 40% of a biller. The software sold to medical practices has not changed this fact, because nearly all of it automates a step inside a job and leaves the job intact. A tool that drafts a note for the physician to finish saves minutes inside an hour the physician still spends. A code suggested for a biller to confirm makes the biller faster at a job the practice still pays a full salary for. Filling part of a prior authorization shortens one task in a position that has to exist either way. In every case the step becomes easier, the salary does not change, and the practice has added the price of software to a payroll it was already paying.

Taking over one whole workflow does more, and it is still not enough, because the workflows are connected. The result of each is the input to the next, so even a workflow performed perfectly on the practice's behalf delivers its result to a person, who uses it to begin the one that follows. Automate the coding completely, and someone still assembles the chart the note was written from, and someone still submits and works the claims built on the codes. Their positions remain, and so does most of the payroll. The payroll decreases only when connected workflows are taken over together, so that the result of each passes into the next with no person between them and nothing is returned to anyone inside the practice to finish by hand. That standard is severe, and anything less than it makes the work more pleasant without making it cheaper. It is also why every tool built for a piece of the work has left the cost unchanged. The tools were built to assist the people, and the cost is the people.

What kept anyone from meeting that standard was the nature of the work itself. Almost every step requires reading and judgment over unstructured language. A fax has to be read before anyone knows whose chart it belongs in. Reconstructing a history means understanding every outside record that mentions the patient. Writing the note means producing prose that a particular physician would recognize as their own. Work of that kind could not be done by software, so every prior attempt automated the structured steps around the human ones and left the people in place. That is no longer true. Software can now read an incoming fax and decide whose chart it belongs in without a person opening it. It can read everything that has accumulated since a patient's last visit, reconstruct what happened in the interval, and write the note in the physician's own style, finished before the patient arrives. This is the change that makes the standard possible to meet.

Meeting the standard has to begin somewhere, and the structure of the work decides where. Almost all of it derives from one document, the clinical note the physician writes for each visit. What the practice is paid depends on what the note contains, because the physician is paid according to what the documentation supports, and a visit recorded at a level below the one actually provided is paid at less than it was worth. What happens after the visit depends on the note as well: the billing codes, the referrals, the orders for tests and medications, and the requests for approval are all taken from it, and the next visit begins where the last note ended. A tool that begins after the note is written is limited by the quality of the note it is given, because it can only act on what the physician already wrote. Improving the note at the moment it is written improves the source, and the improvement passes into every process derived from it. The note is where to begin. To prepare it properly is to reconstruct the full history of what has happened to the patient since the last visit and to write the note in the physician's own style, complete before the patient arrives, so that no one inside the practice has to open the record and assemble the chart. From there, the connected processes are taken over in sequence (the coding, the claims, the referrals, and the rest) until the administrative work that made independent practice unaffordable is no longer performed by anyone the practice employs.

Removing the work has a consequence that should be stated plainly: the cost that disappears is the cost of paying people, so many of these jobs disappear with it. About 680,000 people work as medical secretaries and administrative assistants, and about 195,000 work as medical coders, a job with no patient contact at all. Much of what the 811,000 medical assistants do is work of this kind, and a large share of the billing and coding is performed offshore, in an outsourcing business that passed $34 billion in 2025. For the medical assistant, who is also trained clinically, the change returns the job to that training. For the worker whose entire job is the paperwork, the work ends, and the loss falls on people who did nothing wrong. The question of what happens to those jobs is fair, and the answer depends on what the work was for. This work existed to get the practice paid. It produced no care, and it was paid for by patients in higher prices and by physicians in lost time and lost practices. The jobs that remain are the ones the patient's care depends on directly, and those are the jobs American medicine is short of and will need in greater numbers as the population ages.

The present situation is not permanent, because the cost that produced it can be removed, and once it is removed the decision every physician faces changes. The physician who owns a practice today and cannot reconcile its hours with its income would be able to keep it. The physician finishing training, who chooses employment because opening a practice means first building and paying for a back office, could open one and do only the medicine the training was for. The decline in ownership is the sum of those individual decisions, made one at a time under the same financial condition, and when the condition changes, the decisions change with it. The independent physician, who worked for the patient and for no employer, was the ordinary form of American medicine for nearly the whole of its history. There is no reason it cannot be so again.