EMR integration via FHIR isn't a monolithic problem. Understanding which specific integrations ship cleanly vs. which are still painful shapes realistic project scoping.
Ships cleanly in 2026
1. Patient/Encounter/Observation reads. US Core-conformant reads work well across major EHRs. 2. **SMART on FHIR patient launch. Universal support. 3. Bulk export nightly.** Bulk Data IG is production. 4. CDS Hooks patient-view. CDS Hooks is widely supported. 5. e-Prescribing. MedicationRequest is universal.
Still painful
1. Bidirectional data writes. Reads work; writes back to legacy EHR are custom. 2. Non-Patient scopes. Practitioner and PractitionerRole reads are less well-supported. 3. Custom search parameters. Beyond US Core defaults, coverage varies. 4. Subscription reliability. Subscription delivery isn't always reliable. 5. QuestionnaireResponse write-through. SDC form results back to EHR chart is inconsistent.
Vendor state (mid-2026)
| EHR | Patient read | Bulk export | CDS Hooks | Write-back |
|---|---|---|---|---|
| Epic | Universal | Full | Full | Configurable |
| Cerner Powerchart | Universal | Full | Full | Configurable |
| Athenaclinicals | Universal | Partial | Partial | Custom |
| Meditech | Growing | Partial | Basic | Custom |
Integration scoping
1. Read-heavy integrations → straightforward. 2. Write-back integrations → budget for custom work. 3. Complex workflow integrations → SMART + CDS Hooks + Task. 4. Data warehouse integrations → Bulk data.
Common project scoping mistakes
1. Assuming bidirectional works everywhere. 2. Underestimating vendor-specific auth flows. 3. Overlooking terminology binding differences.
FHIR EMR integration in 2026 is well-defined but not homogeneous. Scope specifically to what your vendors actually support today.
