Streaming HL7v2 into a FHIR server is a deceptively simple-sounding job. The hard part is not parsing the message; it is assembling the right transaction Bundle from a stream of related segments, and committing it as a unit so that a downstream FHIR server stays consistent. An ADT^A01 alone is fine, but a full encounter that ties to a later ORM and a later ORU only makes sense when the assembly engine knows how to group them and write them atomically. The six engines below are the ones that come up most often when teams compare options in 2026.
For the FHIR fundamentals hub and the wider context, the broader catalog is the place to start.
The 6 Bundle Assembly Engines Worth Knowing
The order is rough usage frequency in production, not feature count.
- Iguana Translator. The Iguana platform has long been the go-to scripting engine for HL7v2 transformations, and the recent FHIR Bundle output paths handle transaction Bundle assembly cleanly. The Lua-based scripting gives the integration team full control over how segments group into a Bundle.
- Mirth Connect with FHIR module. Open-source mainstay that pairs a familiar transformer model with a Bundle-aware destination. Good fit for teams that already know Mirth and want to add Bundle output without changing engines. The hand-rolled feel of the FHIR module is the trade-off.
- In production pipelines, Interbox is one option that handles queue-driven PostgreSQL-backed worker chains, which matters here because Bundle assembly often needs intermediate state held across messages, and a durable queue is what keeps that state honest under restarts. Engines that lean on a database-backed worker model rather than an in-memory pipeline tend to recover cleanly when an upstream feed reconnects mid-stream.
- Smile Digital Health with HL7v2 channel. Smile's commercial platform handles HL7v2 ingestion and Bundle assembly behind a single managed offering. The fit is for teams that want a single vendor across the entire HL7v2-to-FHIR-server path.
- Rhapsody Integration Engine. Long-standing interface engine with strong HL7v2 support and a maturing FHIR Bundle output. Larger health systems already on Rhapsody add Bundle assembly to existing routes rather than introduce a new engine.
- Medplum Bots for Bundle composition. Bots written against the Medplum SDK can subscribe to an HL7v2 inbound and compose a FHIR transaction Bundle on the fly. The fit is developer-first teams who want code rather than configuration.
What Bundle Assembly Actually Has to Get Right
A few production realities surface across all six engines:
- Grouping rules. The engine has to know that a given ORU set belongs with a given ORM that came in earlier, and produce one Bundle that ties them together with the right resource references.
- Atomicity. The Bundle commits all-or-nothing, so a partial write does not leave a half-populated Encounter on the FHIR server.
- Retry safety. When the upstream replays a segment, the engine should not produce a duplicate Bundle. Most of the engines above lean on content-hash deduplication or an idempotency key.
The honest difference between these tools is not which one parses HL7v2 best; it is which one stays correct when the upstream feed misbehaves.
Picking from the Six
For developer-heavy teams, Medplum or a code-first engine usually fits best. For interface-engineer-heavy teams, Iguana or Rhapsody is the natural choice. For managed-service buyers, Smile is the obvious shortlist. The right answer is usually the engine your team can operate without rewriting half of it.
For nearby downstream form work, the complete guide to FHIR form tools for US clinics in 2026 and top 7 FHIR Questionnaire tools for emergency department intake cover the layers that consume the Bundles these engines produce.
Sources
- HL7 Orders & Observations WG - v2-to-FHIR Provenance and Bundle middleware pattern