APIs Are the Control Plane, or Pharma Data Pretends to Be Portable

by Marina VasquezHead of Semantic Systems, HMND

Clinical interoperability keeps getting treated like a transport problem. The industry keeps buying cleaner pipes while the real failure sits one layer up, in the identity of the thing being moved, the meaning attached to it, and the permission to reuse it once it lands somewhere else.

FHIR matters because it gives systems an API-shaped way to exchange clinical data, and CDISC matters because it gives submission and research workflows a shared structure for what those data should become. The handoff only works when the patient, specimen, observation, medication, and encounter keep their identity across the boundary, when the vocabularies stay pinned to the same meaning, and when the access rules survive the trip from an EHR to a sponsor platform to a regulatory package.

A FHIR endpoint that serves US Core resources with proper terminology bindings is already doing governance work, whether the team admits it or not. The same is true of a bulk export endpoint that can move longitudinal records out of operational systems, a subscription endpoint that notifies downstream consumers of events, and the implementation guide that says what a lab result, diagnosis, or medication statement actually means in machine terms. The endpoint is the doorway. The contract is the law.

Where reuse succeeds and where it dies

A clinical API can deliver a patient record cleanly and still fail completely for RWE if the source system hides provenance, if the code system mapping is lossy, or if the receiving system cannot tell whether a value came from a clinician, a device, or a claims feed. CDISC has been explicit that RWD and RWE work requires harmonized mapping and terminology conversion, and recent joint FHIR to CDISC guidance exists precisely because raw FHIR resources do not become submission-ready datasets on contact. Anyone still calling that a simple integration problem is decorating the mess.

This is where the pharma and IT overlap stops being a slogan and becomes a very specific set of decisions. The schema contracts, FHIR profiles, USCDI-aligned fields, CDISC SDTM and ADaM targets, define.xml, transformation logs, and terminology services like LOINC, SNOMED CT, and RxNorm decide whether a result stays interpretable after translation. If a sponsor cannot explain how an observation in one system becomes a controlled term in another, the pipeline is theatre. If a manufacturing lot or assay result cannot carry its identifier and provenance into the research layer, the reuse story breaks in the first audit.

The metadata layer does the real work

The metadata layer is where clinical, research, and manufacturing data either gain a second life or lose their history. Provenance, versioning, value set binding, source system, encounter context, sampling time, and transformation trace are the fields that let downstream systems trust the record without rebuilding the story from scratch. That is the difference between a dashboard and an infrastructure. One looks busy. The other can actually mean something.

This is why semantic infrastructure keeps showing up in the serious work. CDISC and related semantic efforts around RWD exist because interoperability fails when disparate sources use different data models and terminologies, even if the transport itself is impeccable. The same problem hits manufacturing when batch lineage, assay metadata, and release decisions sit in one system while clinical or research consumers see only flattened exports. The name on the file changes. The meaning evaporates.

A lab result that moves from an EHR to a sponsor repository to a regulatory submission is a chain of governed reinterpretations. FHIR to CDISC mapping guides, FDA aligned submission discussions, and the growing pressure to define FHIR profiles for regulatory use all point at the same fact: standards only matter when the receiving system can preserve identity, meaning, and permission across the handoff. Otherwise the standard becomes a costume worn by an export job.

Governance decides reuse before code does

Access rules, consent, role based controls, and data use agreements are part of the API surface whether the architecture diagram bothers to mention them or not. If the receiving platform cannot enforce patient level consent, data partitioning, or study specific access, then semantic correctness buys very little because the record cannot be reused safely. Governance is the control plane. Code just obeys.

That is the part teams dislike, because it ruins the comforting fiction that a new interface equals interoperability. It also explains why so many programs stall after the first successful feed. The data arrived. The semantics drifted. The permissions got hand waved. Six months later everyone is arguing about why the RWE model disagrees with the source of truth, and the answer sits in a mapping table nobody wanted to own.

The older I get in this business, the less impressed I am by any system that says it can exchange data and the more interested I am in whether it can preserve who created it, what it meant, and who is allowed to touch it next. FHIR endpoints, CDISC schemas, terminology services, lineage metadata, and policy enforcement belong in the same conversation because pharma never gets to use one without the others.

If this handoff problem is on your desk, hello@example.com is where the conversation starts.