The CCD document quality problem: hidden trust failures in care summary exchange
Most Continuity of Care Documents exchanged between health systems contain structural defects that silently degrade clinical decision-making. Missing medication lists, outdated allergy records, and unmapped terminology codes pass through health information exchanges without any trust validation. The CCD document quality problem is not a format problem; it is a data trust failure hiding in plain sight.
A document that looks complete but isn't
The Continuity of Care Document was supposed to solve the portability problem in healthcare data. A single, standardized XML document carrying a patient's medications, allergies, problems, and procedures from one provider to the next. The reality is different. A 2014 study published in the Journal of the American Medical Informatics Association found that more than 50% of CCDs analyzed contained incomplete or clinically unusable medication data. That number has not meaningfully improved.
The CCD travels well. It arrives on time. It validates against its schema. And it still fails the patient.
What is a CCD continuity of care document?
A CCD is a standardized clinical document based on the HL7 Clinical Document Architecture (CDA) and the ASTM Continuity of Care Record (CCR) specification. It combines structured XML formatting with clinical content sections: medications, allergies, active problems, procedures, immunizations, vital signs, and social history.
The document was designed to give a receiving clinician a snapshot of a patient's medical status at the time of transfer, discharge, or referral. Every certified EHR system in the United States must be capable of generating and consuming CCDs as part of ONC certification requirements under the 21st Century Cures Act.
But generating a CCD and generating a trustworthy CCD are two entirely different things.
Where CCD document quality breaks down
The structural integrity of a CCD can pass every technical validation check and still carry data that harms clinical decisions. The most common failure modes are predictable and measurable.
Medication list decay. CCDs frequently contain medications that were discontinued months or years before the document was generated. EHR systems pull from the active medication list, but "active" in many systems means "never explicitly stopped." A patient transferred from one health system to another may arrive with a CCD listing 23 medications when only 9 are current.
Allergy section defaults. When a patient's allergy status has not been reviewed, many EHR systems populate the CCD allergy section with "No Known Allergies" rather than "Not Reviewed." These are clinically distinct statements. One is an assertion. The other is an admission of ignorance. The CCD treats them identically.
Terminology mapping failures. A CCD generated by Epic may use RxNorm codes for medications. A CCD consumed by Cerner (now Oracle Health) may expect NDC codes. The mapping layer between these terminologies introduces silent errors. A 2018 ONC report found that 30% of medication entries in exchanged CCDs contained terminology mismatches that could alter clinical interpretation.
Problem list inflation. The active problem list in most EHRs accumulates diagnoses over time without systematic reconciliation. A CCD for a 65-year-old patient may list 40+ active problems, including conditions resolved decades ago. The receiving clinician cannot distinguish between a current diagnosis and a historical artifact without manual review.
Missing provenance metadata. Most CCDs do not carry information about when a data element was last verified, who verified it, or what source system originally created it. A blood pressure reading from 2019 looks identical to one from yesterday. A medication added by a specialist looks the same as one added by a patient portal self-report.
The six quality indicators from Crossing the Quality Chasm
The Institute of Medicine's 2001 report Crossing the Quality Chasm identified six aims for healthcare quality improvement: safety, effectiveness, patient-centeredness, timeliness, efficiency, and equity. These six dimensions are frequently cited in healthcare quality discussions, but they apply directly to the CCD problem in ways that are rarely acknowledged.
Safety fails when a CCD carries discontinued medications that a receiving provider prescribes again, causing a drug-drug interaction. Effectiveness fails when a CCD omits a recent lab result that would change a treatment decision. Patient-centeredness fails when a patient's stated preferences and advance directives do not travel with the document. Timeliness fails when a CCD is generated from data that has not been reconciled in months. Efficiency fails when a clinician spends 15 minutes sorting through a 40-problem list to find the 6 that matter. Equity fails when patients who receive care across multiple fragmented systems, disproportionately low-income and minority populations, have the least reliable CCDs.
Every CCD quality failure maps to at least one of these six aims. Most map to several.
What is considered a documentation failure?
A documentation failure occurs when a clinical record is incomplete, inaccurate, outdated, or misleading in a way that could affect patient care or regulatory compliance. In the context of CCDs, documentation failures include omitting active medications, carrying forward resolved diagnoses, failing to record allergy verification status, and transmitting data without provenance metadata.
The Joint Commission identifies documentation failures as a contributing factor in sentinel events. In 2023, communication failures, which include incomplete or inaccurate care summaries, were cited in 59.4% of sentinel event root cause analyses. The CCD is often the primary vehicle for inter-organizational communication during care transitions. When it fails, the communication fails.
A CCD that validates against its XML schema is not necessarily a document that avoids documentation failure. Technical validity and clinical validity are separate questions.
Key statistics
Should I opt out of health information exchange?
This is a question patients increasingly ask, and it deserves a direct answer. Opting out of health information exchange means your CCDs and other care summaries will not be shared between providers through networks like CommonWell, Carequality, or state-level HIEs.
The trade-off is real. If you opt out, you reduce the risk of flawed data propagating across systems. But you also eliminate the possibility of your accurate data reaching a provider who needs it during an emergency. A trauma surgeon at 2 a.m. will not have your medication list, your allergy history, or your surgical history.
The better question is not whether to opt out, but whether health systems should be exchanging documents they have not validated. The current model treats CCD generation as a compliance checkbox. Generate the document. Send it through the exchange. Check the box. Nobody scores the document for accuracy, recency, or completeness before it leaves the building.
Patients should not have to choose between data availability and data accuracy. That is a false choice created by infrastructure that does not include a trust layer.
Why a library of possible answers matters for EHR systems
One of the structural problems underlying CCD quality is that EHR systems often allow free-text entry where coded entries should be required. When a clinician types "penicillin allergy" instead of selecting a coded allergen from a standardized library, the resulting CCD entry may not trigger drug-drug interaction checks in the receiving system.
A library of possible answers, sometimes called a value set or pick list, constrains data entry to standardized, coded options. This approach improves downstream data quality in three specific ways: it ensures terminology consistency across systems, it enables automated clinical decision support in the receiving EHR, and it reduces the reconciliation burden on the receiving clinician.
The medical director's recommendation for EHR configuration typically includes enforcing structured data entry wherever possible, precisely because free-text entries degrade the clinical utility of every downstream document, including the CCD. When a medical director recommends specific EHR configuration changes, they are making a data quality decision that affects every care summary the system generates.
The trust gap no one measures
Here is the core problem: nobody scores a CCD before it leaves one system or after it arrives at another.
Health information exchanges validate XML structure. They confirm the document conforms to the CDA R2 specification. They check that required sections exist. They do not check whether the medication list is current. They do not check whether the allergy section reflects a verified assessment or a system default. They do not check whether problem list entries are clinically active.
This is the care summary data trust gap. A CCD can be structurally valid and clinically dangerous. The document passes every automated check and still misleads the receiving provider.
The SuperTruth Data Trust Index addresses this gap by scoring health data records across eight dimensions: Provenance (25%), Consent (20%), Recency (15%), Quality (10%), Concordance (10%), Validation (10%), Breadth (5%), and Stability (5%). Applied to CCD content, these dimensions expose the failures that schema validation cannot detect.
Provenance scoring reveals whether a medication entry originated from a physician order, a pharmacy claim, or a patient self-report. Recency scoring flags data elements that have not been verified within a clinically meaningful window. Concordance scoring identifies terminology mismatches between source and receiving systems. Quality scoring catches the structural completeness issues that make a CCD technically valid but clinically incomplete.
What continuity of care document integrity actually requires
Fixing CCD document quality requires more than better templates or stricter EHR configuration. It requires a trust layer that operates between generation and consumption.
First, every CCD should carry provenance metadata at the element level, not just the document level. A medication entry should identify its source system, its last verification date, and the role of the person who verified it. This is not a CDA specification limitation; it is an implementation choice that most EHR vendors have not made.
Second, health information exchanges should score incoming CCDs before routing them to receiving systems. A CCD with a DTI score below a defined floor, say 40 out of 100, should trigger a reconciliation workflow rather than being silently deposited into the receiving EHR.
Third, receiving systems should differentiate between verified and unverified data elements within an imported CCD. Instead of merging CCD content directly into the patient's active record, systems should present it in a reconciliation queue with trust indicators for each element. Some EHR systems already support medication reconciliation workflows, but few extend this concept to allergies, problems, or procedures.
Fourth, patients should have visibility into what their CCDs contain before those documents are exchanged. MyBio.Health represents one approach: giving patients a trust-scored view of their own health data so they can identify errors before those errors propagate across systems.
The downstream cost of ignoring CCD quality
When a flawed CCD enters a health system, the cost is not limited to the initial clinical encounter. That data feeds into population health analytics, risk stratification models, quality reporting, and AI training datasets.
A readmission prediction model trained on CCD-derived problem lists will inherit every inflated diagnosis, every resolved condition carried forward, every unmapped terminology code. The model does not know the difference between a current heart failure diagnosis and one that was resolved after valve replacement surgery in 2016. It treats both as active risk factors.
This is how CCD quality failures become AI model failures. The document was the first link in a chain that ends with a clinical decision support alert, a risk score, or a prior authorization decision. Nobody scored the document. Nobody validated the data elements. The model inherited the noise.
The same dynamic affects clinical decision support systems, medication reconciliation, and terminology mapping at every stage of the data lifecycle.
This is a solvable problem
The CCD was a good idea poorly executed. The format itself is not the problem. The absence of a trust layer between document generation and document consumption is the problem.
Health systems that generate CCDs without scoring them are exporting liability. Health systems that consume CCDs without scoring them are importing risk. Health information exchanges that route CCDs without scoring them are propagating both.
The fix is not to abandon the CCD. It is to score every document before it moves. Provenance, recency, concordance, quality. These are measurable dimensions. They can be scored automatically. They can be enforced at the point of exchange.
The DTI Engine scores every health data record 0-100 across 8 trust dimensions before your AI model sees it. If your team is evaluating CCD data for clinical use, population health analytics, or AI training, contact Louis Simeonidis at louis@supertruth.ai or (215) 918-4140.
Further reading:
Jason Alan Snyder
Co-founder of SuperTruth and Artists & Robots, and an inventor on the Data Trust Index patents. Twenty-plus years building technology inside Interpublic Group. He writes here nearly every day on data trust, provenance, and what AI should be allowed to act on, and publishes essays on his Substack.
About SuperTruth · LinkedIn · Substack · jasonalansnyder.com
See it in practice
DTI scores the record, not the patient.
8 dimensions. 0–100. Travels with every record permanently.