- Chronic disease is managed across specialties, but each organization's record system stops at its own walls — fragmentation is the default, not a malfunction.
- Scattered records cost patients repeated tests, conflicting medication lists, and visits spent re-telling their own history.
- Research pays too: when one disease course is split across six systems, no single dataset contains the whole story.
- The 21st Century Cures Act made machine-readable patient access a legal right, which makes the patient the one realistic integration point.
On this page
A rheumatologist at the university hospital. A nephrologist across town. A dermatologist, a cardiologist, a sleep specialist, and a primary-care doctor trying to hold the whole thing together. For one person with lupus, that is six clinicians in four organizations — and six patient portals, each with its own password, each holding a different sixth of her medical life.
Nobody chose this arrangement. That is precisely the problem: nobody chose it.
Fragmentation is the default, not a malfunction
Medical records end up scattered across multiple providers because records belong to encounters, and chronic disease refuses to stay inside one encounter. Every other failure follows from that mismatch.
Each organization licenses its own electronic health record system, configured for its own clinicians and its own billing. When a patient crosses an organizational boundary — a referral, a move, an emergency in another city — the record does not cross with them. A summary might. The trend lines, the reasoning, the half-finished differential usually do not.
Acute medicine can live with this. A broken arm is set and healed inside one institution. Chronic disease cannot: by the World Health Organization's count, noncommunicable diseases now account for roughly three-quarters of deaths worldwide, and the conditions doing most of that damage — diabetes, cardiovascular disease, chronic respiratory disease — are managed over decades, across specialties, through every system a patient ever touches. The record-keeping model and the disease model point in opposite directions.
What six portals actually show you
Each portal shows the slice of you that its institution produced — and presents that slice as if it were the whole. None of them is lying. None of them is right.
Log into all six on the same afternoon and the inconsistencies surface fast. The hospital portal's medication list still includes the drug the nephrologist stopped in March. The allergy lists differ. The primary-care problem list says 'history of' for something a specialist is actively treating. Each system is accurate about what happened inside its own walls; all of them are wrong about you.
| Where the record lives | What it holds | What it rarely shows |
|---|---|---|
| Hospital portal | Admissions, imaging reports, inpatient labs | Care delivered anywhere else |
| Specialist group portal | Visit notes, in-office procedures | Medication changes made by other prescribers |
| Primary-care portal | Problem list, referrals, immunizations | The specialist reasoning behind each plan |
| Lab portal | Result values and reference ranges | The clinical question the test was answering |
| Pharmacy system | Fills and refills | Why a dose changed, or whether it was tolerated |
| The patient | The whole story | Nothing — but in no machine-readable form |
The last row is the one that matters. There is exactly one node connected to every other node in this graph, and it is not a computer.
The costs are concrete
Fragmentation taxes patients in time, money, and safety — and then taxes research on top.
The clinical costs are familiar to anyone who lives this way: the duplicate MRI ordered because last year's images sit in another system's archive; the new-specialist visit that spends fifteen of its twenty minutes reconstructing history; the drug interaction nobody flags because no single prescriber sees the full list. The Agency for Healthcare Research and Quality has spent years documenting how gaps in care coordination produce exactly these failure modes.
The research cost is quieter. A study recruiting from one health system observes only the care delivered there. For a condition managed across many providers, that means truncated timelines, missing treatments, and gaps misread as remission. The patients with the most complex, most informative disease courses — the ones seeing six specialists — are precisely the patients whose data is sliced too thin for any single institution to study. The version of this problem that hits conditions without laboratory biomarkers is even sharper; we cover it in the data gap for invisible illnesses.
The plumbing that now exists
Since the 21st Century Cures Act, US patients have a legal right not just to copies of their records but to machine-readable copies, delivered through standard APIs. The pipes are real; what flows through them is still slices.
The 21st Century Cures Act did two quiet, consequential things. It made information blocking — practices that interfere with access to electronic health information — a violation with penalties attached. And the rulemaking under it requires certified EHR systems to expose a patient-access API built on HL7 FHIR, the same standard the systems use to talk to each other. HIPAA's right of access had guaranteed copies for years; the Cures Act upgraded those copies from paper-shaped to data-shaped.
Honesty requires saying what this has not fixed. The APIs deliver each system's slice; they do not merge slices, reconcile medication lists, or resolve the contradictions between portals. Exchange between institutions works increasingly well for treatment and still lags for nearly everything else — we take the full inventory in EHR interoperability: what actually works.
The integration layer already exists
The realistic fix is not a national merger of record systems. It is equipping the one party already connected to everything: the patient.
Every fix-from-above — regional exchanges, national networks — runs into the same wall: institutions integrate with the institutions they have agreements with. Patients have no such limits. The person is the only entity with the legal right, the standing incentive, and now the technical means to pull from every source they have ever touched. Patient-mediated aggregation is not a workaround; it is the architecture that matches how chronic disease is actually lived. The practical first steps — requesting complete copies, pulling structured data through portal APIs — are laid out in how to download your own health data as FHIR, and the case for treating the assembled result as something valuable is made in your medical paper trail is a research asset.
This is the premise Mithras Health is built on. The platform — pre-launch, covering 28 chronic conditions across 12 categories — treats the patient-assembled record as the unit that matters: records are stored as FHIR R4 with US Core 6.1.0 as the system of record, the patient owns them, and each research study receives only a revocable, scope-limited consent to a defined slice. Aggregation happens once, where the patient is, instead of never, where the institutions are.
The six portals are not going to merge. The institutions behind them have no commercial reason to, and two decades of interoperability policy have improved exchange without producing anything close to one record per person. But the integration layer was never going to be an institution. It is the woman with lupus and six passwords — who has been doing this job, unpaid and unequipped, all along. The fix is to equip her.
Appendix · questions
Questions, answered
Why are my medical records scattered across multiple providers?
Can my doctors just share records with each other?
What is the fastest way to get my complete medical history in one place?
Does fragmented data affect research on chronic disease?
The patient-mediated research-data platform for chronic disease — 28 conditions, worldwide consented sources, one auditable spine. Join the launch list.
The sun is rising on chronic-disease research.
One email when we launch — and nothing else. Patients, researchers, sponsors, and programs all start here.