Observation-based extraction

Observation-based extraction is the simplest of the three extraction mechanisms — it turns coded answers directly into Observation resources, with no mapping setup required. This is the how-to; for the concept and how it compares to the other mechanisms, see Extraction.

How it works

Two things make an item extractable:

1. A code on the item (item.code) — this becomes the Observation's code.
2. The extraction flag turned on (sdc-questionnaire-observationExtract). The flag is inherited down the tree: turning it on for a group opts in every coded question beneath it, and an explicit "off" on one child overrides that inherited "on" for just that item.

A question with several codes can also switch individual codes on or off: a code marked "on" produces an Observation carrying just that code (even if the question itself is not marked), and a code marked "off" is left out of the question's Observation.

The category of the extracted Observations (for example vital signs) comes from the observation-extract-category setting on the questionnaire or an item; the nearest one above an item wins. The builder does not edit this setting, but keeps it when a file is imported and exported.

Every coded item with an answer produces one Observation (status: final). Answers with a questionnaire-unit extension are written as a valueQuantity with that unit; other answer types map to the matching value[x] (string, boolean, integer, decimal, dateTime, Coding).

If a coded question sits inside a repeating group, every instance of the group gets its own Observation (for example, three filled-in rows produce three Observations).

Running it

1. Fill in the form in the preview (or load a QuestionnaireResponse).
2. Open Save ▾ → 🧪 Observations · FHIR JSON Bundle.
3. A small dialog lets you optionally set the subject and author references and the QuestionnaireResponse id that each Observation should point back to. (The Observations are plain FHIR Observation resources; the specification defines no standard SDC profile to stamp on them.)
4. Confirm to download a transaction Bundle of the extracted Observations.

Everything runs in the browser — no server is required.

When to use it vs. Definition-based extraction

Use Observation-based extraction when you simply want each coded answer to become its own Observation — no setup beyond ticking the flag. Use definition-based extraction when you need to populate specific fields of specific resource types instead (e.g. build a Patient and a Condition). Both are described in Extraction.

Next: Definition-based extraction.