'A shoestring and a stethoscope': Atlas MD's Josh Umbehr, M.D., on the real economics of direct primary care

Direct primary care (DPC) has been around long enough that most physicians know the basic pitch: drop the billing, charge a monthly membership fee, keep a smaller panel and spend real time with patients. What takes longer to answer is whether the economics actually work once you are in it: what the ramp-up looks like, what happens when a patient needs a specialist and whether the overhead savings are as significant as the advocates claim.

Josh Umbehr, M.D., started finding out in 2010, when he finished residency and went straight into DPC before the term had much currency in medicine. He’s a family physician and co-founder of Atlas MD, a Wichita-based practice that has since grown into a software platform and consulting operation aimed at helping other physicians make the same move.

Medical Economics spoke with Umbehr about the real math behind the model, what he thinks most physicians still misunderstand about the transition and why he believes the window for traditional insurance-based primary care is closing faster than most people realize.

Interested in DPC? Don’t miss our full guide in Medical Economics Insider: Direct Primary Care 101 — Reclaiming independent practice with less burnout

This interview has been edited for length and clarity.

You were pretty early on direct primary care. How did you arrive at that decision, and when did you know it was the right move?

The short version goes back to around 2000. In undergrad, I started working for a plastic surgeon as his biller and coder. He was an amazing surgeon, but nobody had ever taught him how to bill insurance. You got to see how the sausage was made on that side, and it became clear that it did not really matter how good of a surgeon you were in terms of how you got paid. What mattered was whether you could play the game of coding, modifiers, submissions and all of that.

Growing up, my dad was a trash man. He is a lawyer now, so we still tell people he is a trash man because it is less embarrassing, and he is also our lawyer. But that was a great business. You paid once a month, and we picked up once a week. So I had this sense that business should just be simple, and I had seen a good business run. Health care did not seem like that at all.

That started the journey of finding doctors who were doing anything differently, whether insurance-free fee-for-service or concierge medicine, which was still in its infancy. I watched that movement over the next 10 years and collected the experience of what was working and what was not.

That was our plan all along. We believed the model was broken, and I wrote my personal statement for medical school and residency around this doom-and-gloom idea that family medicine in particular was dying on the vine. It was not supposed to make it past 2010. We just never really believed that. We believed there had to be a better way to do this that would improve things for patients, doctors and the system. So we were all in from an early stage.

For a doctor who is considering direct primary care, what does the financial model actually look like day to day? And what surprised you most about the economics?

I would not say it was a surprise, exactly. What is interesting is probably how easy it is, and that is where people get hung up.

Patients get this very quickly. They hear it the way a consumer hears about a product that is better, faster, cheaper and easier. They are used to people presenting new ideas that way. It is physicians who get more hung up on all the noise they are used to — MACRA, MIPS, insurance regulation, coding, staff and all the rest. It can seem too easy to say, “I just take care of patients.” It feels almost too good to be true.

There is a great example from “The Algorithm” by John McNeil about Tesla getting a car-buying contract down from 44 clicks to one. The point was that people had assumed all 44 of those clicks mattered because that was just how contracts had always been done. In reality, they were 44 ways of saying the same thing. Health care has the same problem. We are not creative enough. We get frustrated at the wrong things and do not fix them. Physicians are so used to being a brick in the wall that they have lost some of the entrepreneurial and innovative instinct on the business side.

The basic math works very well. If you take 600 patients times $50 a month times 12 months, that is $360,000 a year. To a doctor used to the economics of a traditional practice — where you might have five to seven staff, three to five rooms and all kinds of overhead — that number may not sound like enough. But in this model, overhead can be 20% to 25%, you may only need one staff member for one or two doctors and maybe one or two rooms per physician. So you end up keeping the vast majority as salary.

Add another 100 patients or change your price by $10 and that can mean another $60,000 to $70,000. The math is so strong that you do not have to charge a lot to be very successful in this model. I think that is one of the things that could pull students back into primary care.

What is your panel size, and how did you land on that number?

In the beginning, it was a lot of guesswork, but the math pointed us in the right direction.

Concierge medicine used to throw out numbers like 200 to 400 patients compared with the normal 2,000 to 3,000 patient panel. But we knew that math was geared toward a much higher-end version of this model. What we saw from 2000 to 2010 was that many concierge practices struggled to grow. They may have been profitable, but at $25,000 per person per year, that was never going to be a model for the masses.

We also knew 200 was too low. The usual rule is that you see about 1% of your patients in a day. Two patients a day was just not enough. So we tried to maximize the algorithm: more patients means a lower fee and more people covered, but you still have to match it to what a doctor can reasonably do in a day.

In the beginning, the exact number was guesswork, but the math gave us what I think was the best close answer, which was a fair price for a lot of patients.

How do you structure pricing in direct primary care?

Pricing is probably the most complicated part, or at least the most debated, because it depends so much on what your goals are. We never try to tell people exactly what to charge, because there is a lot of sensitivity around that, but we do try to guide them toward some objective truths.

If you want to be in the top 10% of family physicians in terms of income, then you are probably going to need to do top 10% work, be in a rural setting, work in the emergency department or do procedures. We often use hospitals as a reference point because they are charging the most per patient while asking physicians to see the most patients.

What do you want out of the model? Full-time or part-time? A physician partner or not? If you have a partner, your overhead can drop dramatically because you do not need much more space, equipment or staff. That can make the economics even stronger.

For many doctors, the target is somewhere around $250,000 to $300,000 in primary care income, and the math can work very well around that. We often say your patient panel size times what you want to charge, plus about 30% for overhead, gives you some wiggle room.

The trouble comes when practices try to make a full physician income on 100 or 200 patients. It is not impossible, but it is much harder. Speed matters too. If it takes you two years to get full, the lost revenue can take six to eight years to make up. That is why we spend so much time helping doctors understand that they need to recruit aggressively and set prices in a way that lets them grow quickly. It is often the clinics that charge the least that make the most, because it is not just about what you charge. It is about how many patients you can actually attract and retain.

What does lower overhead actually look like in practice? What did you cut, what did you keep and where are the real savings?

There is a great idea in “The Lean Startup” that in the beginning, all you really need is a shoestring and a stethoscope. You are the product.

If I only had $10,000 to spend, I would probably buy an EKG machine, a pill counter and a vitals machine. The rest is a little optional. Even if you start with no staff, patients often love that because everything runs through you. If the phone rings, you answer it. That may sound intimidating to physicians, but patients are so used to phone trees, bad customer service and not seeing the same people that the fewer steps you have between you and them, the better.

As the practice grows, it becomes a reasonable-size office, usually 700 to 1,000 square feet per doctor, one full-time RN, MA or LPN, one exam room per doctor and the basics like phone, EMR and malpractice coverage.

You do not need a billing department. You do not need to deal with MACRA or MIPS reporting. You are not chasing the next Medicare rule change. Most EMRs in this space automate what little billing there is, so you spend maybe 5% of your time on administrative work and 95% on patient care or just living your life.

That overhead ends up being really marginal. We spend more on coffee in the waiting room than we do on EKGs. That is a good example of putting hospitality back into health care, but doing it in a very efficient way.

What are the biggest challenges direct primary care physicians still face when their patients need specialists, hospitalizations or other outside care?

In a lot of ways, some parts are much better, and other parts are still just the larger system being the larger system.

In this model, you have more time with patients. I had a VIP patient once ask me, “Are you a good doctor?” and I joked, “No.” What I meant was that the point of the model is not that I have to be some kind of superhero. In the old model, a good doctor had to see you in seven minutes, make you feel heard, ask all the right questions, make the diagnosis, explain the treatment, document it and get paid. In this model, I can spend an hour with the patient, talk to a specialist afterward, email the patient back and now even use tools like OpenEvidence and other artificial intelligence (AI) tools to explore the issue more deeply.

That means fewer referrals in general. Keeping a patient out of a broken system is often the best thing we can do. We also have access to cheaper meds, better lab pricing and doctor-to-doctor telemedicine consults. Dermatology is a great example. I can get a dermatologist consult for $30 in a day or two. It would cost me more than that in staff time just to navigate the local referral process, where the patient might wait six months.

But when a patient really does need to go into the system, they are no worse off than anyone else. If they are paying cash, ironically they can sometimes be a better customer in the current system. Imaging is a good example. If you need a CT for back pain and we are trying to go through insurance, that might take one or two weeks. If we are paying cash, I can often get you in today.

If we do need to send someone out, we do our best to guide them, send over records, call the practice, get them on a wait list and help them understand what to expect. That is also where this starts to overlap with employers, because they are coming to us too and saying, “How do we solve this?” They want better insurance, lower cost and still good access to primary care, urgent care, specialty care and emergency care. DPC gives us the time and ability to help on all of those fronts.

For a primary care physician who is burned out in traditional practice, what is the honest timeline and financial runway they need before making the jump to direct primary care?

There is a best-case scenario and then a more normal scenario.

If you are converting an existing practice panel — say 2,000 patients — and you set your pricing correctly, we routinely see doctors recruit 200 to 300 patients in the six to eight months before the switch. Those are the best patients you will ever recruit, because they already know you and you already know them. Everybody else you have to go out and market to.

If you are starting from residency or moving to a new location, then you do not have that advantage. In those cases, good growth is about 40 to 50 patients a month. If you are doing that, you can be full and profitable in six months and at your full doctor income in around 12 months. That is really good.

The dangerous zone is more like 10 to 20 patients a month. Doctors can fool themselves into thinking that is okay, but if you map it out, you are not full for two or two and a half years. That is where it gets deceiving. You need to see steady growth, and again, it is often the practices that charge the least that end up making the most.

The broader point is that not every doctor wants to practice the same way. Some want part-time, some want a bigger panel, some are good at technology, some are not. Some want to be employees of a growing DPC group rather than start their own. If they join an already successful practice with a wait list or strong recruiting engine, the transition can be much faster. We are not fully at the point where that transition is instant for most doctors, but I think it is heading that way.

What role is AI playing in direct primary care?

I love it. I am a technophile by default, and I think it is really interesting.

In some ways, I joke that it is not actually that special. It is in the same family as spell-check or Grammarly. The phrase I heard a few years ago that still holds up is that artificial intelligence decreases the work of work. That is exactly where physicians are going to embrace it, because it is taking over the parts of the job that are not helping. It is not taking away the most valuable parts, which are the communication, the availability, the relationship and the time with the patient.

Note-taking itself does not add value. The hour-long conversation where you are focused and not typing or trying to juggle what to remember, that is the high-value part. The tool is only as good as the data you put into it, but what we are seeing with OpenEvidence and similar tools is a significant drop in referral usage because what used to be a difficult specialty question can now produce an excellent answer quickly, and I can share that with the patient.

I really can see a future where family medicine, pediatrics and internal medicine are powered by AI in a way that lets us handle much more of the low-lying fruit. I think dermatology could become a little more like radiology, where specialists are reviewing a lot of material quickly and guiding the physician on what to biopsy, how to follow up and what to do next.

So I think we will actually get more care while doing a lot less annoying work. The real problem in the current system is that it often pits artificial intelligence against physicians. If the tool helps you chart faster, the system just makes you see more patients. Or it uses the AI-generated note to try to bill more or block more. That does not help the patient. But when you move into a primary care model that is not driven by insurance friction, the whole lens shifts. Then the question becomes: how do we make doctors as effective as possible at taking care of patients? That is where AI becomes really exciting.

Where do you think physicians still have the biggest misconceptions about what direct primary care actually requires?

Physicians are very risk-averse. We deal with risk around patients all day long, and we are not trained in business. So one misconception is that you somehow have to be great at business to make this work. Another is the idea that society will judge you because now you are supposedly focused on money.

To me, the branding needs to be the opposite. This is the most patient-focused care you can get. If we take “do no harm” seriously, then that should include financial harm. We are doing financial harm to patients in a broken system where they are paying premiums that are more than their mortgage to get metformin that costs pennies a pill.

Another misconception is that DPC is still basically concierge medicine for wealthy people, or that cash-based care must mean bad medicine — that if a patient is paying cash, then you are just there to give them what they want. It is true that bad medicine has always been profitable, and bad medicine is always going to be popular with some people. But that is not what makes this model work. The successful practices are successful because they have a relationship, and because they can tell patients no when no is the right answer.

I tell patients that I am the right-answer machine. If it is a urinary tract infection, antibiotics are usually the right answer. If it is a cold, they almost never are. The point is that you can trust me to give you the right answer even if it is not what you wanted to hear. That kind of professionalism and confidence is what makes the model work, not cutting corners or doing bad medicine.

How has the patient relationship changed for you in this model?

There is no barrier between you and the patient. There is no nurse knocking on the door every seven minutes. There is no insurance company to blame or hide behind. In a weird way, it is much more raw.

It is an hour or more if they need it, explaining the care plan, answering questions and then being available afterward by call, text or email. You cannot just rest on your laurels and say, “Well, that is what I told you in the office.” The relationship continues.

I think that raises the standard of care. And it also creates a more honest dynamic where you can say, “I do not know,” or “That is a great question,” or “We did not do those labs before because we were trying to be mindful of costs, but now we have more reason to get them.” Patients are bracing themselves for a reason not to trust the system, and you can often turn that around just through transparency.

The simplest moneyball stat is that the more time patients get with you, the more likely they are to trust you. Seven minutes is not enough. Ten staff members between you and the patient is a problem. When there are so few steps and so much transparency, trust becomes the default.

Where do you see direct primary care heading over the next five years?

I have probably been guilty since 2010 of saying we are just around the corner from peak insurance, but I believe it now more than ever.

The system is going to break, and when it breaks, I think it will happen pretty quickly. We are already seeing people in 2026 dropping insurance midcycle because they just do not want to pay for it anymore or cannot afford it. Once they do that, they are instantly hundreds or thousands of dollars richer every month. They go from cash-poor and insurance-rich to the opposite, and they become excellent direct-pay customers. They want value, they want transparency and they will push the model to become the best version of itself.

So I really do think that within three to five years, insurance-free primary care will be the default or at least the assumed model for a lot more people. Maybe that sounds bullish, but that is what I expect.

Is there anything else physicians should keep in mind?

I think one really interesting area will be how doctors grow, evolve and connect these models, especially in how they work with employers. We have moved mountains already, but we are still at the foot of the Rockies. There is a lot more to move.

As soon as the game is no longer “How much money can you squeeze out of insurance?” but instead “How good of a business model do you have?” I think a lot of things will change. That could include hospital-at-home models, smaller physician-led hospitals or outpatient emergency departments.

Between artificial intelligence making the barriers to innovation much smaller and the business model leaning more toward insurance-free care by default, I think we will probably see more change in health care in the next five or 10 years than we have in the last 20 or 30.