Edit this page

FHIR PEWS one-pager: how to read a Bundle without losing the will to live¶

Date: 2026-04-21 Audience: digital-pews contributors who are new to FHIR

1) What is a Bundle?¶

A FHIR Bundle is a container that groups related resources.

For a PEWS event, one Bundle might include:

• Patient
• Encounter
• Several Observation resources (vitals)
• One score Observation (NPEWS total, optionally sub-scores)
• Provenance (who recorded it, when)

Mental model:

• Bundle = envelope
• entry[] = list of documents inside the envelope
• entry[i].resource = the actual document

2) Where is the clinical data?¶

Usually in Observation resources.

The minimum fields to look at first:

1. resourceType (should be Observation)
2. code (what was measured)
3. effectiveDateTime (when measured)
4. value[x] (the result)
5. dataAbsentReason (if skipped/missing)
6. component[] (if there are sub-parts)

Most common result locations:

• valueQuantity.value: numeric result (for example HR 132)
• valueCodeableConcept: coded result (for example AVPU = A)
• component[].valueQuantity or component[].valueCodeableConcept: grouped results

3) PEWS vs NPEWS1¶

• PEWS: generic clinical concept (paediatric early warning score)
• NPEWS1: a specific National PEWS profile/example naming used in NHS Clinical Observations work

In practice:

• PEWS is the umbrella term
• NPEWS1 is one concrete model/profile instance used in examples

4) A tiny observation example with annotations¶

{
  "resourceType": "Observation",
  "status": "final",
  "code": {
    "coding": [
      {
        "system": "http://loinc.org",
        "code": "9279-1",
        "display": "Respiratory rate"
      }
    ]
  },
  "subject": { "reference": "Patient/123" },
  "encounter": { "reference": "Encounter/456" },
  "effectiveDateTime": "2026-03-18T02:00:00Z",
  "valueQuantity": {
    "value": 32,
    "unit": "/min",
    "system": "http://unitsofmeasure.org",
    "code": "/min"
  }
}

How to read this in 10 seconds:

• What? respiratory rate
• When? 2026-03-18T02:00:00Z
• Value? 32 per minute
• Who/where? linked by subject and encounter references

5) Why the nesting exists¶

FHIR is verbose because it carries:

• clinical meaning (code systems)
• units and comparability
• provenance and traceability
• references between resources
• safe handling of missing data

That complexity helps interoperability, but it is painful to read at first.

6) Fast triage method for any PEWS Bundle¶

When opening a large Bundle JSON, do this in order:

1. Confirm Bundle.type and count entry[]
2. Find all entry[].resource where resourceType = Observation
3. For each Observation, scan code -> effectiveDateTime -> value[x]
4. Separate vital signs from score observations
5. Check for missing/skipped fields via dataAbsentReason

If you only do those five steps, you can understand most payloads quickly.

7) Mapping to this repo's chart model¶

Chart model fields in this repo are mostly direct equivalents of Observation values:

• respiratoryRate <- Observation(valueQuantity)
• oxygenSaturation <- Observation(valueQuantity)
• heartRate <- Observation(valueQuantity)
• bloodPressureSystolic/Diastolic <- Observation(component[])
• avpu <- Observation(valueCodeableConcept)
• temperature <- Observation(valueQuantity)

Score fields:

• pewsTotal <- score Observation (value or components)
• escalationLevel <- interpretation or extension/profile rule

8) Common confusion points¶

1. "I cannot find the data"

2. You are probably looking at Bundle/metadata instead of entry[].resource Observation payloads.

3. "Everything is ids and profiles"

4. Ignore identifiers and meta first; read code/effective/value first.

5. "Why are there multiple observations with similar timestamps?"

6. One timestamped round often emits many Observation resources, one per parameter.

9) Practical next step¶

Take one real PEWS example and produce a compact extraction table:

• time
• parameter code/display
• value
• unit
• skip reason

That table is the easiest bridge between FHIR payloads and UI chart rows.