Clinical and EHR Data

Clinical data from electronic health records is the mirror image of claims: everything claims are not. It is clinically rich, current, and true to what actually happened to the patient, and it is trapped inside organizational walls. Understanding both properties is essential to building analytics that see the whole patient.

What clinical data offers

Where claims show billing, the EHR shows medicine:

• Lab values, not just that a test was ordered but its result.
• Vital signs, blood pressure, weight, and the like.
• Medications, what was prescribed and, sometimes, filled.
• Problem lists and diagnoses as the clinician recorded them.
• Clinical notes, the narrative detail no coded field captures.

And it is current: clinical data reflects care as it happens, without the claims lag. For anything that needs to be timely or clinically precise, blood pressure control, a lab-based measure, a real-time alert, the EHR is the source.

Structured versus unstructured

A defining analytic challenge is that clinical data comes in two very different forms:

• Structured data is coded and queryable: lab results with values, medications with codes, vitals with numbers. Increasingly this is exposed through FHIR resources, the standard the interoperability module covers.
• Unstructured data is free text: the clinician’s notes, where much of the real clinical story lives but which is hard to compute on. Extracting it requires natural-language processing and remains imperfect.

The walls problem

Worth remembering: an EHR sees only the care its own organization delivered. Everything the patient does elsewhere, the emergency visit across town, the specialist on a different system, the prescription filled at another pharmacy, is invisible to it. This is the exact inverse of claims: the EHR is deep and narrow where claims are broad and shallow. Neither alone shows the whole patient, which is why serious value-based analytics must combine them, and why the interoperability that moves clinical data across walls matters so much.

Working with clinical data

Building on EHR data means extracting it (increasingly via FHIR APIs), normalizing it across systems that record the same thing differently, mapping local codes to standards, and reconciling it against claims. The payoff is the clinical precision and timeliness claims lack. The cost is fragmentation: the more organizations a population touches, the more clinical sources must be stitched together, which the next module addresses.

Key takeaways

• EHR data is clinically rich and current: results, vitals, medications, and notes, without the claims lag.
• It comes as structured (coded, increasingly via FHIR) and unstructured (free-text) data, the latter hard to compute on.
• Its defining limit is the walls: it sees only its own organization’s care, the inverse of claims, so the two must be combined.