DPC Directory

The panel-size critique of DPC: what holds up and what doesn't

DPC Directory Editorial

The short answer: half of the critique holds up, and half has not been shown yet. The supply point is arithmetic: a clinician who looks after a few hundred patients looks after fewer people than one who carries a traditional panel, so if that clinician would otherwise have carried the larger panel, the system has less capacity. The disparity point, that direct primary care (DPC) "increase[s] disparities in access to care," may turn out to be true, but the study cited for it did not measure who gains or loses care, and did not separate DPC from concierge medicine.

Here is what the National Academies said about direct primary care panel size, what its sources show, and what we could not find studied yet.

What the National Academies report says about DPC

In early November 2026 the National Academies of Sciences, Engineering, and Medicine released Ensuring Access to High-Quality Primary Care in the United States, a follow-up to its 2021 primary care report (news release, dated November 5). Its main recommendations are a "common fund" for primary care, more total investment in primary care, and a larger primary care workforce.

DPC gets one paragraph, on page 39 of the prepublication copy, under "Additional changes in the primary care landscape." The committee calls DPC "one prominent market response" to problems with financing, workforce supply, practice capacity and administrative burden. Citing the Direct Primary Care Alliance, it says DPC clinicians "care for far fewer patients than non-DPC clinicians (average panel sizes of 500–600 versus 1,700, respectively)." It adds that "DPC practices are popular with patients who can afford it and with clinicians who value autonomy, less administrative hassle, and smaller patient panels," and that membership grew by more than 800 percent from 2017 to 2025 (citing Hint Health), "but this still remains a small proportion of primary care nationally."

The paragraph ends with the sentence this post is about:

"DPC does, however, have a major system-level drawback: the low panel sizes reduce the effective supply of primary care clinicians available to serve the overall population and increase disparities in access to care (Zhu et al., 2025)."

The reference list (page 87) identifies Zhu et al., 2025 as "Growth in number of practices and clinicians participating in concierge and direct primary care, 2018-23," Health Affairs 44(12):1473–81. The pages are marked "Uncorrected Proofs," so wording and page numbers may change in the final edition.

The study the sentence cites

Jane M. Zhu (Oregon Health & Science University) and co-authors from Johns Hopkins and Harvard built a list of concierge and DPC clinicians from a public online directory, FindMyDirectDoctor, matched them to federal provider records and a commercial database (IQVIA OneKey), and traced them back year by year to 2018. We read the full text of the author manuscript on PubMed Central; the journal version is behind Health Affairs' paywall, and we did not see it.

What the study found:

  • Concierge and DPC practice sites in its sample grew by 83.1 percent from 2018 to 2023, and the clinicians in them by 78.4 percent.
  • Independent ownership fell from 84.0 percent of practices to 59.7 percent, while practices tied to corporate owners grew by 576 percent.
  • Advanced practice clinicians (nurse practitioners and physician assistants) rose from 32.7 to 40.3 percent of the clinicians in these practices.
  • Nearly 30 percent of physicians entering these practices came from health systems or integrated delivery networks. "[R]elatively low rates" came from federally qualified health centers, which the authors say may reflect barriers to leaving or "the relative durability of the safety-net workforce." Many advanced practice clinicians who entered had no prior employer.

What it did not do, in the authors' own words:

  • It did not separate DPC from concierge. "Given that some practices may self-identify as either concierge or direct primary care, we did not distinguish between these models." About 60 percent of the clinicians billed Medicare, "suggesting concierge or hybrid practice," and about 40 percent did not, "presumably in direct primary care." The authors note that concierge retainers average "a few thousand and up to tens of thousands of dollars" a year on top of insurance billing, while DPC "is often framed as a model focused on increasing access and affordability."
  • It did not measure panel sizes. Smaller panels appear in the paper as a shared feature of both models, cited to earlier work, not as a finding.
  • It did not look at patients. There are no data on who the practices serve, what patients pay, or whether anyone lost access to care.
  • It treats access as an open question. The paper's last sentence: "the impact of concierge and direct primary care models on access to care, quality of care, and the primary care workforce will be vital for patients, clinicians, and policy makers to understand." Asked about access in an interview with the National Institute for Health Care Management (NIHCM) Foundation, which helped fund the study, Zhu said these models "may also improve access and satisfaction for enrolled patients," and that "their implications for system-wide access remain uncertain and likely vary by model and market."

The authors' concern cuts both ways. They write that the growth in advanced practice clinicians "may help alleviate physician shortages," and they frame the question as whether these models can "mitigate broader primary care shortages instead of compounding them." The study is a count from a public directory that the authors say likely understates the number of practices. It raises the access question; it does not answer it.

What holds up: the supply math

Access is already limited: the National Academies' news release, citing an annual scorecard, says nearly 3 in 10 adults lack a usual source of care. As an illustration only, using the report's own figures: if one clinician leaves a traditional panel of 1,700 for a DPC panel of 550, roughly 1,150 patients need another clinician. That is a real cost. Smaller panels are not a side effect of DPC but part of its appeal, and the report itself says clinicians value them.

Two caveats about the numbers, which the report does not mention.

The 500–600 and 1,700 figures are not quite what the cited source says. The committee cites the Direct Primary Care Alliance's State of Direct Primary Care report. The Alliance is a membership and advocacy group for DPC physicians; its report, published in July 2026, comes from a survey emailed to its members and shared in DPC Facebook groups in October and November 2024. We could not find a 1,700 figure in it, or an average panel of 500–600. What it reports is that most respondents' panels were "well below the often-cited word-of-mouth benchmark of 600 patients," that 68.4 percent said their panels were not yet full, and that among physicians with full panels, sizes "cluster more closely around the 400–700 range." Its average monthly fee, about $98, does match the National Academies report. The report does not say where 1,700 comes from, and it is not in the Alliance report we read.

A separate American Academy of Family Physicians (AAFP) data brief, from a 2026 survey of family physicians who had engaged with DPC content, put the average panel in 2025 at 504, up from 402 in 2024. That average covers all 313 respondents (a 9.3 percent response rate), of whom 53 percent were practicing in a DPC setting. Neither survey is a random sample.

The supply cost depends on where the clinician would otherwise be. The arithmetic assumes every DPC clinician would otherwise carry a full traditional panel. What we know about origins mostly points that way: in the AAFP survey, 63 percent opened a DPC practice after leaving a health system or group, and in Zhu's data nearly 30 percent of entering physicians came from health systems. That supports the direction of the supply concern, though not its size. For some clinicians, the alternative may have been leaving medicine, retiring early or cutting hours; in the AAFP survey, 8 percent went straight from residency, and in Zhu's data many advanced practice clinicians had no prior employer. We found no study that measures how many fall into each group.

There is also an open question on the other side of the ledger: whether DPC patients use urgent care and emergency rooms less, which would free capacity elsewhere. We have not found peer-reviewed evidence that settles this either way, so we are not counting it.

What has not been shown: the disparity claim

"Increase disparities in access to care" is a claim about which people end up with less care. To show it you would need to follow patients: who left a clinician's panel when the clinician converted to DPC, whether those patients found another clinician, and how long it took, by income, insurance and place. The study the report cites does none of that, and its lead author calls the system-wide effect "uncertain." The sentence states as a finding something the cited study leaves as an open question.

That does not make the concern baseless, and a fair reading should say so:

  • The report notes that DPC is "popular with patients who can afford it." A monthly fee is a barrier for some people, and memberships are hard to combine with Medicaid, because a Medicaid provider must accept the Medicaid payment as payment in full for covered services (42 CFR 447.15; our Medicaid FAQ).
  • In the AAFP survey, physicians estimated that on average 80 percent of their patients had private health insurance. That sample includes hybrid and concierge practices, and the brief gives no comparison group, so the figure does not tell us how DPC patients differ from anyone else.
  • Zhu's sample shows growing corporate ownership and hybrid models, which the authors say "do not fully align with traditional definitions" of DPC. What happens to access if those models grow is not known.

On price: DPC fees sit well below concierge retainers, and in the DPC Alliance survey rural practices charged less on average ($81.56 a month) than urban ones ($110.00), but a lower price says nothing about who is served or turned away. Plausible is not demonstrated: the evidence the report cites for the disparity effect is a study that did not test it.

What we see in the directory

We are a directory, not a research team, so none of this settles the question. But it is what we can show without guessing:

  • Prices. Our DPC membership prices by state report draws on the listed practices that publish a price (of the 2,455 we list). It shows a median headline monthly fee of $99 and a median adult (age 40) price of $95 a month, on data as of November 22, 2026. Medians vary a good deal by state, and the report says how each figure is built. Our methods page explains how we source and check listings. These prices say nothing about who is served or who is turned away.
  • Medicaid. One owner's description shows how murky the line can be. Dr. Jillian Klaucke, whose rural practice is in its eighth year, said on the My DPC Story podcast: "we partner with Medicaid, though I have opted out of Medicaid" (episode page, September 13, 2026). That passage does not say how Medicaid patients are served at the practice.
  • New panels start small. Dr. Brian Juan, who opened Metronome Family Medicine at age 50, said on the same podcast that the practice had gone from 110 to 114 members before the recording that morning; the episode has a chapter on growing the panel and setting boundaries (episode page, September 20, 2026). That is one new practice early in its growth, not evidence about mature panel sizes, but it fits the DPC Alliance finding that most panels surveyed were not yet full.
  • DPC is not concierge medicine. The difference matters here, because the cited study counted both together. Our DPC vs concierge FAQ sets out how they differ.

The AAFP's CEO and the common fund

A week after the report came out, R. Shawn Martin, executive vice president and CEO of the AAFP, spoke on a panel at an open meeting of the National Academies' Standing Committee on Primary Care on November 12. According to Medical Economics, Martin said: "If we're really serious about primary care as a foundation for the health care system … we've got to get really comfortable with being uncomfortable about insurance," and "I don't know that primary care needs to sit inside an insurance structure in the traditional way that we've talked about here today." Medical Economics reports that Martin called the common fund the right idea but not necessarily ready, given how complex health regulation and payment are, and later clarified that people should still have health insurance.

Be careful reading that as an endorsement of DPC. Medical Economics does not report Martin mentioning direct primary care. The common fund he was discussing is the report's proposal: a fund, created by states (or by Congress as an alternative), that would pool payments from public and private insurers, pay primary care practices standardized rates, and remove patient cost-sharing, whatever a person's insurance. That is a different mechanism from a DPC membership, where each patient pays a practice directly. The common fund is not a DPC membership. What the two ideas share is a view that primary care may need to be paid differently from standard fee-for-service claims.

What we still don't know

  • Where DPC clinicians would otherwise be: in a full traditional panel, a part-time job, or out of medicine. The size of the supply effect depends on it.
  • What happens to patients when a practice converts: who leaves, where they go, and how long it takes them to find care.
  • How DPC patients differ from everyone else by income, insurance and place, from data rather than physician estimates.
  • DPC on its own, separated from concierge and hybrid models, as the Zhu authors themselves suggest.

Until those studies exist, the fair summary is the one at the top: fewer patients per clinician is a real trade-off, and whether DPC widens gaps in who gets care is still an open question. If you are weighing DPC for yourself, you can browse practices by state and compare what each one publishes.

This is general information about health policy research, not medical, legal or financial advice. Quotes from the National Academies report are from the prepublication copy, marked "Uncorrected Proofs," and may change in the final edition. Sources were read on November 13–14, 2026. DPC Insider, a DPC trade news site, also pointed out the gap between the report's sentence and the cited study; we checked the report and the study ourselves.

Questions

Do small DPC panels reduce the primary care supply?

Per clinician, yes. A DPC clinician with a panel of a few hundred patients looks after fewer people than one carrying a traditional panel, so if that clinician would otherwise have carried the larger panel, the system has less capacity. How big the effect is depends on where DPC clinicians would otherwise be working: in a full traditional panel, part time, or out of medicine. We found no study that measures it.

Does DPC increase disparities in access to care?

Not shown yet. The National Academies report says it does, citing Zhu et al. (Health Affairs, 2025). That study counted concierge and DPC practices together, did not measure panel sizes, and had no data on which patients gained or lost care. Its lead author has said the effect on system-wide access remains uncertain. The concern is plausible, since memberships cost money and are hard to combine with Medicaid, and in an American Academy of Family Physicians survey physicians estimated that on average 80 percent of their patients had private insurance. But the study cited does not test it.

Where do the 500 to 600 and 1,700 panel figures come from?

The National Academies report attributes both to the Direct Primary Care Alliance's 2026 State of Direct Primary Care report. We could not find either figure stated there. That report, based on an October to November 2024 survey of DPC physicians, says most panels were below a commonly cited benchmark of 600 patients and that full panels clustered around 400 to 700. It gives no figure for traditional panels.