Interoperability in healthcare is the ability of different information systems to exchange data and use it meaningfully. The second half of that definition carries the weight. Transmitting a document from one hospital to another is straightforward; ensuring the receiving system understands that a particular code means a particular diagnosis, recorded at a particular time, about a particular patient, and can act on it computationally, is the actual problem.

Poor interoperability has direct clinical consequences: tests repeated because prior results are unavailable, medication errors because a full list cannot be assembled, delayed care while records are requested, and research and public health surveillance limited by fragmented data.

The Levels of Interoperability

  • Foundational. Systems can transmit and receive data. The receiving system need not interpret it.
  • Structural. The format and syntax are defined, so the receiving system can parse the message and locate fields reliably.
  • Semantic. Both systems share the meaning of the content through common terminologies and models, so data can be used computationally rather than merely displayed.
  • Organisational. Governance, policy, trust frameworks, consent arrangements, and workflow allow exchange to happen in practice between institutions.

Many systems achieve structural interoperability and stop there. A PDF discharge summary transmitted successfully is readable by a human but cannot populate a medication list, trigger a decision support rule, or contribute to a registry. Semantic interoperability is what makes data actionable.

The Core Standards

HL7 and FHIR

Health Level Seven International develops the most widely used healthcare exchange standards. HL7 version 2 messaging remains extensively deployed for laboratory results, admissions, and orders, and continues to carry a large share of routine traffic. Clinical Document Architecture defines structured clinical documents.

FHIR — Fast Healthcare Interoperability Resources — is the modern standard. It defines discrete resources such as Patient, Observation, Condition, MedicationRequest, and Encounter, accessed through RESTful web APIs using common web technologies. This modularity allows an application to request precisely the data it needs rather than exchanging whole documents, and it lowers the barrier for developers outside traditional health IT. FHIR supports profiles and implementation guides that constrain the base specification for a particular national or use-case context, which is necessary for real interoperability but also means that FHIR conformance alone does not guarantee two systems will interoperate.

Clinical Terminologies

  • SNOMED CT is a comprehensive clinical terminology providing coded concepts with defined relationships, used to record diagnoses, findings, procedures, and other clinical content in a computable way.
  • LOINC identifies laboratory tests, measurements, and observations, allowing a result from one laboratory to be recognised as the same test elsewhere.
  • ICD, maintained by the World Health Organization, classifies diseases and causes of death for statistics, reporting, and in many systems reimbursement. It is a classification rather than a clinical terminology, and serves a different purpose from SNOMED CT.
  • Medication terminologies such as RxNorm and ATC support unambiguous identification of medicinal products and substances.
  • DICOM defines the format and exchange protocol for medical images and associated metadata.

Terminology mapping between these systems is routine but lossy: converting a detailed clinical concept into a broader classification code discards specificity that cannot be recovered.

How Data Are Exchanged in Practice

  • Point-to-point interfaces connect two systems directly. Simple individually, unmanageable at scale, since each new participant multiplies the number of interfaces required.
  • Integration engines sit between systems, translating formats and routing messages, and remain the backbone of most hospital environments.
  • Health information exchanges operate as regional or national intermediaries, holding or brokering records so participating organisations can retrieve information about shared patients.
  • APIs allow applications to query data directly, subject to authorisation. FHIR-based APIs underpin patient-facing access and third-party application ecosystems.
  • Patient-mediated exchange gives patients access to their own records and the ability to share them, shifting some of the burden of aggregation to the individual.
  • Federated analytics allows queries to run against data held locally, returning results rather than moving records, which supports research and surveillance while limiting data movement.

Persistent Barriers

Patient Identity Matching

Linking records requires confidence that two records describe the same person. Where a unique national health identifier exists, this is comparatively tractable. Where it does not, systems rely on probabilistic matching using name, date of birth, address, and other attributes, which produces both failed matches and, more dangerously, incorrect matches that merge two people's records.

Data Quality and Variation

Clinical data are recorded for care and administration, not for exchange or analysis. Free text predominates in many areas, coding practice varies between institutions and countries, and local code sets are common. Structured data may be incomplete because entry is burdensome.

Incentives and Information Blocking

Sharing data has often been commercially or competitively disadvantageous for both vendors and provider organisations. Regulatory responses have made practices that unreasonably impede access, exchange, or use of electronic health information subject to sanction in some jurisdictions, and have mandated standardised API access.

Legal and Governance Complexity

Data protection frameworks permit sharing for care and, subject to conditions, for research and public health. In practice, uncertainty about lawful basis, differing institutional interpretations, and cross-border complexity slow exchange. Cautious misinterpretation of privacy law is a documented obstacle in its own right.

Legacy Systems and Cost

Health systems run long-lived software with limited modern interface capability, and replacing or extending it is expensive and operationally disruptive.

Privacy, Security, and Secondary Use

Health data are among the most sensitive personal data. Governance rests on several requirements: a lawful basis for processing; purpose limitation and data minimisation; role-based access control with auditing of who accessed what; encryption in transit and at rest; transparency to patients about use; and rights of access and correction.

Secondary use — research, service planning, quality improvement, and public health surveillance — raises additional questions. De-identification reduces but does not eliminate re-identification risk, particularly with rich longitudinal or genomic data, which is why trusted research environments and secure data environments have become the preferred model: analysts work with data inside a controlled environment rather than receiving extracts. Frameworks such as the European Health Data Space aim to establish common rules for both primary and secondary use across jurisdictions.

Security is a patient safety issue as well as a privacy one. Ransomware and other cyber incidents have caused significant disruption to clinical services, which is why resilience, backup, and continuity planning belong in interoperability programmes rather than alongside them.

Why Interoperability Matters Clinically

  • Safety. Complete medication and allergy information at the point of prescribing reduces avoidable harm.
  • Efficiency. Access to prior results and imaging avoids duplicate testing and associated cost, delay, and radiation exposure.
  • Continuity. Care spanning primary, secondary, community, and social care depends on information following the patient.
  • Emergency care. Clinicians treating an unfamiliar patient benefit from access to essential history.
  • Research and surveillance. Population-scale linked data support outcome research, pharmacovigilance, and outbreak detection.
  • Patient agency. Access to one's own records supports informed participation in care.

Sources

  • HL7 International — FHIR specification and implementation guidance
  • SNOMED International — SNOMED CT clinical terminology
  • Regenstrief Institute — LOINC
  • World Health Organization — International Classification of Diseases
  • DICOM Standard — medical imaging exchange
  • European Commission — European Health Data Space; General Data Protection Regulation
  • Office of the National Coordinator for Health Information Technology — interoperability and information blocking rules
  • Organisation for Economic Co-operation and Development — health data governance recommendations