FHIR and HL7 are necessary but not sufficient. The hard parts are identifier matching, result timing and who owns the record of truth.
Every HMIS procurement promises interoperability. Few deliver it, and the failures are rarely about the standard. HL7 and FHIR are well specified. What breaks is the operational detail around them.
Identifier matching is the first wall
Two systems that each hold a correct patient list will still disagree unless they agree on identity. Without a deliberate matching strategy — national identifier first, then a composite fallback with manual review — you create duplicate records that clinicians learn to distrust.
Results need a defined owner
When a laboratory result is published, one system must be authoritative and the other must be a consumer. Ambiguity here produces the worst outcome: a clinician looking at a stale value and believing it is current.
Sequence the interfaces, do not parallelise them
Attempting registration, laboratory and radiology interfaces simultaneously spreads attention across three failure modes. Sequencing them — each stabilised in production before the next begins — costs less overall and produces far fewer incidents.
Interoperability is an operational commitment, not a protocol checkbox. Budget for the matching logic and the reconciliation review, not just the message format.
Written by the delivery team. Figures referenced are from customer deployments and measured against each customer's own pre-deployment baseline. They are not a guarantee of results in a different operating environment.