Belgian Cancer Registry on FHIR - Local Development build (v0.1.0) built by the FHIR (HL7® FHIR® Standard) Build Tools. See the Directory of published versions

Data flow

Data flow

The BCR data ecosystem is fed by two parallel streams with different sources, granularities, and submission channels. The published research dataset is a derived view over both, with pseudonymisation and aggregation applied.

Stream 1 — Clinical registration (one row per cancer case)

Hospital MOC discusses a patient
        │
        ▼
Coordinating physician fills in Bijlage 55
        │
        ├──── Section 1-11 (universal)              ──► BCRCancerRegistrationForm
        ├──── Vervolg 1 (only if breast tumour)     ──► BCRBreastTumourSupplement
        ├──── Vervolg 2 (MOC attestation)           ──► BCRMOCForm
        └──── Vervolg 3 (subsequent follow-up MOCs) ──► BCRFollowUpForm
        │
        ▼
Submission via WBCR
        │
        ├── manual entry (web UI), OR
        └── batch extraction from HIS
        │
        ▼
BCR internal database

• Who submits: hospitals with oncological care programmes (legally obliged).
• Submission window: within 4–6 months of incidence date.
• Channel: WBCR — manual web entry or HIS batch.
• Authentication: eHealth portal (eID / itsme).
• Patient identity: INSZ / NISS via eHealth IdentifyPerson lookup.

Stream 2 — Pathology screening submissions (one row per lab sample)

Pathology / clinical biology / haematology lab
        │
        │ Processes a screening sample (breast / cervix / colorectal)
        │
        ▼
Structured dataset + pathology report
        │
        ▼
sFTP transfer to BCR
        │
        ├── monthly cadence (cervical samples)
        └── quarterly cadence (breast, colorectal, general)
        │
        ▼
CHP (cyto-histopathology) register      ──► BCRChpSample
                                            ├── BCRChpBreastSample
                                            ├── BCRChpCervixSample
                                            └── BCRChpColonSample

• Who submits: every Belgian pathological anatomy / clinical biology / haematology lab (legally obliged since 2025 to use sFTP exclusively).
• Codes used: ICD-O-3 morphology + topography + CODAP-2017 (BCR-specific lesion-type codes).
• Granularity: one row per sample, not per patient. Same patient and same tumour can appear across multiple rows.

Stream 3 — Published research dataset (derived)

BCR internal database (streams 1 + 2 + registry-wide processing)
        │
        │ Pseudonymisation
        │ Field derivation (numeric ICD-10, age categories, vital-status dates)
        │ Aggregation (totaltum, multiple, borsttelling)
        │ Filtering to non-identifiable variables
        │
        ▼
"Cancer in Belgium" research dataset      ──► BCRCancerCase

• What's in it: ~38 variables, see BCRCancerCase.
• What's NOT in it: patient name, INSZ, treatment chronology, MOC details, all Tier 2 disease-specific extensions, all Tier 3 NIHDI project data.
• Released to: external researchers and institutions on formal data request.

Why the model split matters

The three streams are deliberately modelled as separate logical structures rather than a single unified model:

Trying to use a single resource for all three would either lose information from streams 1 and 2 or pollute the research-dataset shape with form-level concerns. Separation is honest.