Edit this page
Digital PEWS — specification & decisions
This folder is the source of record for what Digital PEWS must do and why it is
built the way it is. Every fact here is traceable to a document in
../reference-sources/ (the NHS SPOT/NPEWS specifications, the
published age-band charts, and the configuration/decision-pathway PDFs) or to the code in
../chart/ (the chart component) and ../demo/ (the demo harness).
New here? Read ../README.md for how to run the app, then
decisions.md for the shape of the project, then the reference
specs below.
Reference specifications (transcribed from the SPOT-NPEWS .xlsx)
These are faithful transcriptions of the national specification. Must requirements are
non-negotiable; requirement IDs (U3.10, C3.17, …) are load-bearing — don't renumber them.
| Document |
What it covers |
spot-npews-ui-spec.md |
NHS SPOT/NPEWS UI specification (MOSCOW-prioritised U* requirements). |
spot-npews-spec.md |
NHS SPOT/NPEWS clinical/technical specification (C* requirements). |
Scoring (JSON is the source of truth)
| Document |
What it covers |
npews-scoring-spec.json |
Canonical numeric scoring spec — every threshold, age-band bound and escalation mapping. Edit this, then run npm run generate:scoring. Drift-tested. |
npews-scoring-tables.generated.md |
Human-readable unified reference table. Generated from the JSON — do not hand-edit. |
npews-scoring.md |
Clinical narrative only: the per-age-band respiratory-distress descriptors and implementer notes. |
Clinical policy & data
| Document |
What it covers |
escalation.md |
Escalation levels, non-score triggers, "think sepsis", and the ISBAR aide-memoire. |
ed-pews.md |
ED source intake, initial differences, unresolved policy and gated implementation plan; not implemented ED support. |
../reference-sources/README.md |
Provenance and hashes for the four public ED PDFs, alongside the unchanged inpatient baseline. |
data-model.md |
The two component inputs — Patient and Observation (raw vitals) — plus data-entry reference codes. |
fhir.md |
The FHIR adapter contract (as implemented) and the conformance-testing approach. |
patient-derived-data-assessment.md |
Aggregate analysis of the gitignored ED resource, disclosure-risk decision, synthetic replacement strategy, and controlled local-testing constraints. |
../SAFETY.md |
Current clinical safety status, initial hazard log, controls, transferred responsibilities, and open assurance work. |
Project decisions & direction
| Document |
What it covers |
product-boundary.md |
The component-only product boundary, responsibility allocation, compliance claims, and deliberate C2.2 non-conformance. |
age-band-boundary-policy.md |
The open clinical policy question at age-band boundaries, current behaviour, hazard implications, and governance route. |
requirements-traceability.md |
Requirement-by-requirement allocation and evidence for the component critical path; remaining source requirements are explicitly queued for allocation. |
decisions.md |
Durable decision log (ADR-style, newest first). Start here to understand the project's choices. |
bug-log.md |
Open findings awaiting review, including scorer behaviours pinned by characterisation tests but not clinically endorsed. |
implementation-notes.md |
Clarifications discovered while building, plus a manual visual-QA checklist. |
react.md |
The component-architecture decision: framework-neutral Web Component vs React. |
oracle-cerner-registration.md |
How to register the SMART app with Oracle Health (Cerner) code Console, and the separate per-tenant go-live process a customer must initiate. |
roadmap.md |
Forward-looking work (including Web Component Phase 2). |
releases.md |
0.x development version policy, single-command release cascade, prerequisites and recovery. |
web-component-phase2-spec.md |
Acceptance criteria for isolation, multiple instances, lifecycle, types, and distribution. |
visual-regression-testing-plan.md |
Browser interaction, screenshot, baseline-governance, and visual-CI plan. |
Conventions
- The JSON scoring spec wins. Where any prose disagrees with
npews-scoring-spec.json, the JSON is correct (and a drift
test will be red). Regenerate artifacts with npm run generate:scoring.
- Ground claims in a source. Prefer citing a
reference-sources/ document or a code
path over asserting a fact. If a requirement can't be traced, mark it as an
implementation decision in decisions.md.
- Clinical safety over features. Visual and numeric fidelity to the national NPEWS
charts takes precedence — see the hard constraints in
../CLAUDE.md.
- Record decisions once. New durable decisions go in
decisions.md,
not scattered through the specs.