Every chart mixes two kinds of information. Some of it sits in fields the software can sort, count and report on. The rest is written in sentences, or arrives as a scanned page, and only a person can read it well. Knowing which is which decides how much of the chart you can reuse later.
This guide is for physicians, clinical staff and practice administrators who set up templates or decide how notes get written. It compares structured and unstructured data in an electronic health record (EHR), gives examples of each, and maps where each one belongs in a typical outpatient chart. For a deeper look at the standards behind structured fields, see our guide to structured data in an EMR.
What is the difference between structured and unstructured data in an EHR?
Structured data is recorded in defined fields with a known format, such as a date, a number with a unit, or a code from a standard list. Unstructured data is free text, dictation, scanned documents or images that do not follow a fixed data model. Software can sort and count the first kind directly. The second kind has to be read.
The two are not rivals. A structured field records that something happened. Narrative records why, how the patient described it and what the clinician was thinking. A useful chart keeps both and makes clear which one to trust for which question.
| Structured data | Unstructured data | |
|---|---|---|
| Form | Fields, pick lists, numeric values, codes | Free-text notes, dictation, scanned letters, images |
| Examples | Vital signs, medication and dose, allergy, diagnosis code, appointment date, a pain score | History of present illness, assessment reasoning, an outside physician's letter, a patient's own account of an accident |
| Good for | Search, trends over time, reports, alerts, exchange with other systems | Context, nuance, uncertainty, the clinician's reasoning |
| Weak at | Explaining why; anything the field designer did not anticipate | Counting, comparing visits, building a report without rereading |
| Main risk | A filled field that is wrong or out of date looks just as certain as a correct one | Important facts get buried and are missed on review |
What are common examples of each type?
Most items with a single correct value are captured as structured data: demographics, vital signs, medications, allergies, diagnoses, lab values and appointment dates. Narrative is used for history, reasoning and anything that needs explaining. Scanned outside records and images are unstructured unless someone abstracts values from them.
The United States Core Data for Interoperability (USCDI), maintained by the Office of the National Coordinator for Health IT (ONC), is a useful reference. It is a standardized set of data elements, grouped into data classes such as patient demographics, that serves as a baseline for exchanging health information between systems. If an item appears in USCDI, it is a strong candidate for a structured field in your chart.
Structured fields work best when they use a shared vocabulary. The National Library of Medicine describes standardized terminologies as organized sets of concepts, each with a unique code, so different systems can mean the same thing by the same entry. Familiar examples include ICD-10 (International Classification of Diseases, 10th revision) for diagnoses, CPT (Current Procedural Terminology) for procedures, LOINC (Logical Observation Identifiers Names and Codes) for lab tests and RxNorm for medications.
Where does each type belong in the chart?
Use a structured field for anything you will need to compare across visits, report on, or send to another system. Use narrative for anything that needs reasoning or context. Many sections of a note work best as a mix: fields for the facts, a short free-text box for the explanation.
| Chart section | Best captured as | Why |
|---|---|---|
| Demographics and insurance or case details | Structured | Used for scheduling, billing and matching records. Errors here spread everywhere. |
| Chief complaint and history of present illness | Mostly narrative, with a few fields (onset date, mechanism of injury) | The story matters, but key dates should be findable. |
| Medications and allergies | Structured | Needed for prescribing checks and reconciliation. |
| Vital signs, pain scores, range of motion | Structured, using the same scale each visit | These are the values you will trend over time. |
| Exam findings | Mixed | Fields for measurable findings; narrative for anything unusual. |
| Assessment and plan | Coded diagnosis plus narrative reasoning | The code supports billing and reporting; the reasoning supports the decision. |
| Outside records and imaging reports | Unstructured attachment, with key values abstracted if needed | The original must stay intact; abstracted values make it usable. |
A common mistake is to turn everything into checkboxes. A note built only from pick lists can read as identical from visit to visit, which makes it harder, not easier, for a later reader to see what changed. Keep a free-text box wherever the clinician might need to explain.
