FHIR data strategies in production converge on seven patterns. Understanding them shapes both design and operations.
1. FHIR-native storage. Resources as canonical form. No proprietary + facade duplication.
2. $validate in write path. `$validate` operation at every write. Reject on error.
3. Terminology binding at write time. Every coded field validated against ValueSet.
4. Reference integrity monitoring. Nightly checks that references resolve.
5. Bundle transactions for related writes. Related resources atomic.
6. Bulk data for analytics. Bulk Data IG $export feeds warehouses.
7. Metrics per resource type. Prometheus dashboards.
Data quality metrics that predict outcomes
| Metric | Healthy | Warning | Critical |
|---|---|---|---|
$validate pass rate |
>97% | 95-97% | <95% |
| Reference integrity | >99% | 97-99% | <97% |
| Duplicate rate | <2% | 2-5% | >5% |
| Terminology compliance | >98% | 95-98% | <95% |
Investment areas
1. Data governance team (1-2 FTE). 2. Automated monitoring pipelines. 3. Terminology infrastructure. 4. Bulk data pipeline. 5. Observability tooling.
Common strategy mistakes
1. No data governance ownership. 2. Manual $validate review. 3. Unversioned terminology. 4. No reference integrity checks. 5. Bulk data as afterthought.
Governance workflow
1. Weekly data quality metrics review. 2. Terminology update cycle (quarterly SNOMED, weekly RxNorm). 3. Reference integrity audits. 4. Duplicate detection cadence. 5. Audit trail completeness.
FHIR data strategies compound over years. Sites investing in the seven patterns above see data quality improve; sites that don't see degradation.
