June 24, 2026
There is a version of this argument that sounds reasonable. Electronic health records were immature in the early 2000s. Adoption was uneven. Systems didn't talk to one another. The solution, the logic goes, is more adoption, better interoperability standards, and continued investment in the same infrastructure.
The United States has spent twenty years and hundreds of billions of dollars testing this hypothesis.
The results are in. Fragmentation has gotten worse. And the reason has nothing to do with the pace of adoption or the quality of the technology. It has to do with mathematics, and with a structural reality that no EHR vendor has ever been designed to address.
The Adoption Era That Made Things Worse
The Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009 was the largest coordinated push for EHR adoption in the history of American healthcare. By the early 2020s, EHR adoption among hospitals and physician practices had reached levels that would have been unimaginable two decades prior.
Over that same period, a research team led by Barnett and colleagues (2021) tracked what actually happened to care fragmentation among Medicare beneficiaries. Their finding inverts the adoption thesis entirely:
Outpatient care became measurably more fragmented over the two decades of highest EHR investment in history.
Patients were seeing more providers. Those providers were working across more practices. The coordination web had expanded faster than any documentation system could track it.
More EHRs did not produce more coordination. They produced more documented fragmentation.
This is not a failure of execution. It is a failure of category. The hypothesis that documentation infrastructure solves a coordination problem was always asking the wrong tool to do the wrong job.
The Mathematical Reality
To understand why, consider what coordination actually requires at the clinical level.
A study by Pham and colleagues (2009) examined the peer physician networks of primary care physicians treating Medicare patients. The finding reframes the entire coordination conversation:
A single primary care physician manages relationships with an average of 229 peer physicians across 117 different practices.
This is not an outlier. This is the median structural condition of primary care in the United States.
Now consider what this means for a super-high utilizer patient, the population that consumes a disproportionate share of behavioral health resources and most urgently requires coordinated care. These patients see a median of seven different physicians from four different practices in a given year (Pham et al., 2009).
When that patient's primary care physician needs to coordinate care, they are not navigating a manageable clinical network. They are attempting to maintain meaningful coordination relationships with 229 colleagues across 117 practices: simultaneously, continuously, without any shared infrastructure to support it.
This is not a communication problem. It is not a motivation problem. It is a combinatorial problem. The number of point-to-point connections required to coordinate care across 117 practices does not scale with effort or goodwill. It scales geometrically. And it collapses under its own weight long before any individual clinician runs out of determination.
No EHR solves this. An EHR that documents every encounter at every one of those 117 practices with perfect fidelity still requires someone, a care manager, a clinician, or a coordinator, to manually traverse that network, locate the relevant information, synthesize it into a usable clinical picture, and act on it in time to matter.
The filing cabinet got better. The number of filing cabinets got larger. The person responsible for reading all of them did not get more hours in the day.
The Communication Gap That Technology Did Not Close
The Pham and Barnett data describes the structural condition. A separate body of research documents the clinical consequence.
Timmins and colleagues (2022) examined communication patterns between primary care physicians and specialists across a nationally representative sample. Their finding is one of the most quietly damning in the fragmentation literature: communication gaps between primary care and specialist physicians are decade-unchanged despite the proliferation of EHR adoption across that same period.
Referring physicians were not receiving consultation notes. Specialists were not receiving the clinical context they needed from referring physicians before appointments. The basic coordination loop: send patient, receive report, integrate findings, adjust treatment: was failing at rates that had not meaningfully improved in ten years of technology investment.
This is not because clinicians stopped trying. It is because the technology investment was directed at documentation, and the problem was never documentation. The problem was the absence of a shared coordination layer: a Common Point of Care that could actively connect providers across institutional boundaries rather than passively storing their individual records in isolated systems.
The Patient Cannot Fill the Gap
When infrastructure fails, the burden transfers somewhere. In behavioral health, it transfers to the patient.
Kern and colleagues (2019) documented this directly: patients with serious mental illness have no reliable mechanism to put their treating providers in touch with one another. They cannot verify which providers have access to their records. They cannot initiate a clinical communication between their psychiatrist and their primary care physician. They become, by default, the coordination infrastructure that the system failed to build.
This is not a minor inconvenience. For patients managing serious mental illness, often simultaneously navigating cognitive symptoms, medication side effects, trauma histories, and substance use. The demand that they personally maintain the coherence of their own clinical narrative across seven providers and four practices is not a reasonable ask. It is a patient safety failure dressed as a care model.
When the patient cannot carry the full picture accurately, and they cannot, because no patient should be expected to, The gaps that result are not documentation gaps. They are safety gaps. A provider acting on an incomplete history because the patient could not remember a prior medication trial is not making an informed clinical decision. They are making the best decision available given a structurally impoverished information environment.
What More Adoption Will Not Fix
The argument for continued investment in EHR adoption rests on the premise that the problem is incomplete implementation. If every provider were on the same system, with the same standards, with the same interoperability protocols, coordination would follow.
The Barnett and Pham data dismantle this premise at the structural level. The number of peer physician relationships a primary care physician must manage is not a function of EHR adoption rates. It is a function of how American healthcare is organized: in independent practices, across competing health systems, under different regulatory frameworks, with different financial incentives. EHR adoption does not change that organizational reality. It documents it more efficiently.
The coordination problem requires a coordination solution. Not a documentation solution with better interoperability. Not a health information exchange that makes it slightly easier to request records from another institution. A dedicated infrastructure layer that actively synthesizes the longitudinal clinical narrative across every system a patient has touched, and makes that synthesis available to every authorized provider, in real time, at the point of care.
Until that infrastructure exists, the coordination web will continue to expand. The documentation will continue to improve. And the gap between what the record contains and what the clinician actually knows, at the moment they need to make a safe decision, will continue to produce the harm that better filing systems were never designed to prevent.
The health information exchange represents the natural next proposal: if EHR adoption within systems is insufficient, then cross-system connectivity through HIEs and interoperability frameworks could close the remaining gap. The evidence is equally precise. As of 2024, only 19 percent of behavioral health and substance use treatment facilities participate in a regional or state HIE, and 67 percent of non-participating facilities report being either unfamiliar with HIEs or unaware of whether one exists in their service area (Chang and Owusu-Mensah, 2026). And even for programs that do participate, the HIE solves a routing problem: moving records from one system to another. It does not solve the synthesis problem. The distinction between routing records and synthesizing them into an actionable longitudinal clinical picture is examined in the first installment in this series and will be addressed in depth in a later installment on interoperability infrastructure.
___
This is the second in a series examining the assumptions that are preventing behavioral health organizations from solving their most consequential patient safety problem.
This is the first in a developing series examining the assumptions that are preventing behavioral health organizations from solving their most consequential patient safety problem.
Next: why fragmentation is not a care quality issue: the four-link clinical chain that connects a missing psychiatric record to a preventable death.
Featherglass Health builds clinical memory infrastructure for high-acuity behavioral health organizations. Download our clinical intelligence brief, Beyond the AI Scribe, to read the full evidence base behind this series.
___
References:
Pham HH et al. Primary care physicians' links to other physicians through Medicare patients: the scope of care coordination. Ann Intern Med. 2009;150(4):236–242.
Barnett ML et al. Trends in outpatient care for Medicare beneficiaries and implications for primary care, 2000 to 2019. Ann Intern Med. 2021;174(12):1658–1665.
Timmins L et al. Communication gaps persist between primary care and specialist physicians. Ann Fam Med. 2022;20(4):343–347.
Kern LM et al. Patients' and providers' views on causes and consequences of healthcare fragmentation in the ambulatory setting. J Gen Intern Med. 2019;34(6):899–907.