Master Patient Index in FHIR: Architecture Choices That Compound

Master Patient Index in FHIR: Architecture Choices That Compound

Master Patient Index in FHIR: Architecture Choices That Compound

MPI architecture choices made at project start shape MPI operations for years. Three architectural patterns cover most FHIR-based deployments.

Architecture 1: Embedded MPI in FHIR server.

MPI logic lives in the FHIR server itself. Every Patient POST triggers matching logic. Aidbox MDM module works this way. Best for single-vendor stacks.

Architecture 2: Standalone MPI with FHIR sync.

Dedicated MPI service (Verato, NextGate) sits alongside the FHIR server. Patient resources sync between them. FHIR server queries MPI for match candidates. Best for multi-source integrations.

Architecture 3: MPI as gateway.

All Patient CRUD goes through MPI as a gateway. MPI decides whether to create new or reference existing Patient. Downstream FHIR server sees only resolved identities. Best for high-volume ingestion pipelines.

Trade-off comparison

Aspect Embedded Standalone sync Gateway
Ops simplicity High Medium Low
Multi-source support Low High High
Match sophistication Basic Advanced Advanced
Vendor lock-in High Low Medium
Merge tooling Basic Advanced Advanced

Integration with FHIR primitives

All three architectures use Patient.link to represent merges. Downstream systems must follow the link chain for correctness.

Common architecture mistakes

1. Embedded MPI where multi-source integration was needed → migration to standalone. 2. Standalone MPI without gateway → duplicate creation during eventual sync lag. 3. Gateway MPI without proper caching → point-of-care latency spikes.

Vendor state (mid-2026)

MPI Embedded fit Standalone Gateway
Verato Add-on Native Native
NextGate Add-on Native Configurable
Aidbox MDM Native Add-on Add-on
MITRE FRIL Standalone only Yes Custom

MPI architecture is a 5-10 year decision. Get the right pattern for your source-count and volume trajectory.

Emily Tran

HIM specialist from San Diego. Covers clinical document exchange, C-CDA, and the long tail of EHR migration projects.