Data interchange standards in healthcare IT--computable semantic interoperability: now possible but still difficult, do we really need a better mousetrap?

From Clinfowiki
Jump to: navigation, search

This is a review of Charles Mead's 2006 article, “Data Interchange Standards in Healthcare IT-Computable Semantic Interoperability: Now Possible but Still Difficult, Do We Really Need a Better Mousetrap? Journal of Health Information Management, 20: 1: 71-78.

Article Review

This article is an interesting and foundational piece in differentiating HL7 v 2.x and HL7 v.3. The move to HL7 v. 3 by its stakeholders is due to the need for more depth, breadth, and complex health care data information exchanges. The minimum data set for HL7 v. 3 is known as computable semantic interoperability (CSI). [1]


CSI is defined as unambiguous information exchange. Semantics is defined as meaning whereas syntax is known as structure. The fundamental building blocks of semantics are datatypes. Dataypes can be specified when its formal meaning and its legal set of computational operations can be performed on an instance of the datatype that are rigorously specified. Healthcare requires complex coded terms within coding terminologies. An example of this is a coding system name, version, primary and alternate code.

Interoperability is the exchange of information or data by humans or machine. The exchange of structured data is syntactic interoperability. An example of machine-to-machine syntactic interoperability is web pages built by HTML or XML. A syntactic or properly structured page can be read by any machine, with the meaning to be attached by the human viewer.

Semantic or human interoperability relies on meaning that is unambiguously exchanged between humans. The medical record contains documents such as consults, progress notes and others that rely on a specific set of vocabularies and common practices to pass meaning from clinician to clinician. CSI also requires that each machine that processes data makes its decisions based on the same meaning.

Healthcare data is docu-centric, or collected in terms of a document like a history and physical, progress note, etc. Health care data is extracted from documents, which is integrated with data from documents, or other non-documents. Documents are considered to be permanent, versus messages which are transient.

Clinical Document Architecture (CSA) is part of the “HL7 Toolkit”. The beauty of CSA is that it is derived from v.3 Reference Information Model (RIM), which non-document message structures are derived from. In a simple description, this means that CDA documents possess computationally semantic interoperability with data that is obtained via non-document v.3 messaging sources.

When a process is scalable, it works for all instances. Semantic scalability occurs when a given reference and message is unambiguously understood by an increasing number of systems on a plug-and-play basis.

Reference Information Model (RIM) is the reason behind the HL7 v.3. The limitations behind v 2.x include: that v2.x could not meet the robust need for semantically scalable data exchange in a cost-effective manner. The strengths of v.3 contrast the weaknesses of v.2.x. One of the differences is that v 2.x lacked a common information model that spans all domains of interest. Version 3, on the other hand, spans clinical, administrative and financial data, and covers unambiguous definitions of the common structures in health care data information exchanges. These structures form the core of XML tag data set for a given v.3 message. RIM has become an ANSI standard.

Another difference is that version 2 lacked a robust infrastructure for specifying and binding concept based terminology values to specific message elements. Version 3 allows RIM to be interwoven with many terminologies such as SNOWMED, LOINC, DICOM, and MIAME/MAGE. Version two lacked a message development process that was top-down. Version three, on the other hand, provides a number of tools to assist in building RIM-conforming interchange structures for ANSI balloting.

The HL7 v.3 RIM defines the semantics of a common set of financial, clinical and administrative data structures. RIM defines a high level backbone that contains five abstract structural concepts:

Entity: things including organization, person, non-person, material and living subjects. Role: time-based-capability, capacity or competency. Participation: Role in the context of an act. Act: financial, clinical or administrative definitions, plans, occurrences. Act relationship: the semantics between acts.

A number of attributes specifies each one of the backbone classes. By binding to an HL7 v.3 datatype, the semantics of each attribute is specified. The semantics of a data exchange structure are the combination of pre-defined semantics expanded or modified by virtue of binding to codes or data values specified by HL7 or external organizations, i.e., domain-specific terminologies.

The author stated that the goals of the National Health Information Infrastructure of the Department of Health and Human Services (HHS) can be used to plead the case for HL7 v. 3. The NHI seeks to improve patient care by making clinical knowledge via decision support available to clinicians at the point of care delivery embedded in the EHR. Semantically robust processing will be required for determining the right information available to clinicians at the right time. Another NHI goal is to connect clinicians by cataloging, machine sorting and other processing.

A group of leading technology companies formed the Interoperability Consortium in response to the HHS’s call for recommendations on how to achieve the NHI goals. Group members include Oracle, Microsoft, and IBM, which demonstrated the need for cooperation among traditionally rigid vendor boundaries. The Consortium recommended the following:

  1. Interface definitions for health information service providers and NHIN-provided services.
  2. Service-level requirements.
  3. Data exchange standards, including syntactic defining data interchange structures and methods, semantic standards, defining data meaning with sufficient robustness, so data can be understood by all processing machines.HL7 v.3 with it’s associated RIM, and other pertinent industry standards, provide the basis for semantic standards.


The author concluded that the Consortium’s recommendation for HL7 v. 3 clearly demonstrates it’s understanding of the complex need in health care IT for a data exchange that encompasses CSI. The author also stated that anything short of CSI as a solution, even if it is faster, less costly and easier to implement in the short run, will not meet the goals of the Office of the National Coordinator’s (for health care IT) for transforming the nation’s health care delivery system.

Related papers


  1. Mead CN. Data interchange standards in healthcare IT--computable semantic interoperability: now possible but still difficult, do we really need a better mousetrap?