Part 1 of this series introduced the role of standard terminologies in meaningful use (MU). Part 2 illustrates some of the challenges encountered to support the implementation of MU.
Challenge #1: Many terminologies are needed to support the EHR and MU
Currently, no one terminology or classification system contains everything that is needed for the EHR, so encoding patient data for MU requires multiple standards. The Office of the National Coordinator for Health Information Technology (ONC) has adopted an initial set of vocabulary standards to support the proposed requirements of MU, shown in Table 1.
Table 1: MU standard terminologies (HITSC Implementation Workgroup, 2011)
|MU objective||Adopted code set|
|Demographics||ISO 639-1, OMB|
|Problem list||SNOMED CT®, ICD-9-CM*|
|Clinical summaries||SNOMED CT, ICD-9-CM*, LOINC®, RxNORM|
|View and download personal health information (PHI)||SNOMED, ICD-9-CM*, LOINC, RxNORM|
|Lab test results||LOINC|
*Transition to ICD-10 after October 1, 2013
Challenge #2: Implementing standards
Another challenge is the characteristics and complexities of coding systems themselves as well as the lack of a recommended, comprehensive implementation strategy that an organization can follow. These characteristics present both implementation and management challenges:
- Heterogeneity: Standards have content that differs in areas of focus (laboratory versus pharmacy), granularity (level of detail), organization, and even digital file format (text delimited, spreadsheet, databases, etc.).
- Content changes: Semantic “drift” or “shift” both signify a change in the meaning of a code—the identifier for a concept in a terminology. Semantic drift describes a change that happens gradually over time, while a semantic shift is a significant change in meaning at a single point in time. Moreover, codes can be re-used or deleted, and the result is encoded data that can no longer be interpreted.
- Versions: Standards have different formats and releases, and the changes in each subsequent version have to be reconciled with the content of previous versions, potentially impacting encoded data. For example, if a standard code is removed by the standard’s developing organization, the data is no longer interpretable.
- Coverage: In many cases, a standard terminology or coding system cannot provide all the content needed to encode data in a target domain. Reasons include changing medical knowledge and events; content that is truly local in nature, such as locally compounded medications; granularity differences between the data collected and the concepts available in the terminology; immature standards, etc. Consequently, there is always data collected for which there is no standard code available or identified in the MU measures.
- Historical compatibility: Today, many health information systems use local and proprietary internal codes or free text to collect and store patient data. This data is not interoperable with newer data encoded using standard terminologies. Often how the local codes are created and maintained and how systems use them have implications: for example, if local codes are hard-coded into the systems, replacing the terminology requires rewriting the software. Ironically, in some situations, an organization can achieve semantic interoperability with the outside world, but lack interoperability within its own legacy data and systems.
Challenge #3: Managing local or proprietary codes
Today most EHR systems use local, vendor-specific, and proprietary internal codes or free text to collect and store patient data. These local terminologies consist of widely variable codes that have not been mapped to standard terminologies, and they also lack the framework for integrating with other health information systems. Often these types of local interface terminologies pose the types of problems shown in Table 2.
As mentioned earlier, much depends on how the local terminology was created and how it is maintained, as well as with how an organization’s information systems are using the terminology. Local codes are often stored within tables referred to as master files, which are referenced by applications to encode and decode data. The codes may be hard-coded into the systems, so replacing the terminology requires rewriting the software.
Another layer of complexity arises when an organization begins to implement standard terminologies and then must determine how to maintain backward compatibility with data stored using a local terminology.
Table 2: Obstacles to managing local and proprietary terminologies
|Not concept-based||Various valid and invalid terms are being used to represent the same concept||Dyspnea vs. shortness of breath vs. disnea vs. SOB|
|Code removal or re-use||Codes for deleted or inactivated terms or concepts are reassigned to new terms||Prior to July 2002 NDC code: 00074433501 = LiposynPost July 2002, NDC code: 00094433501 = Paclitaxel|
|Lack of version control||No rigorous mechanism for versioning or standard maintenance protocols||Ad hoc updates|
|Ambiguous content||Unclear description and no formal definition||“Blue” in a specimen domain|
Challenge #4: Creating an implementation plan
Without an established roadmap for standards implementation, organizations must ask themselves:
- What data must be encoded using standards?
- Which standard should we use for each type of data?
- Can we create a link between local terminologies and the standards through mappings or can we use the standards directly?
- How do we maintain and establish governance policy with regard to this content?
Once these questions have been answered, each organization must evaluate the following with regard to their implementation plan:
- Interoperability with legacy data and systems: Is it necessary to maintain compatibility with data previously stored using non-standard codes or systems that produce or depend on non-standard codes?
- Cost/effort to implement standards: Different approaches will require very different costs and effort to implement. Are the expected immediate or long-term benefits of the approach worth the additional cost?
- Cost/effort to stay up-to-date and maintain mappings: If you decide to map between local and standard terminologies, what are the maintenance requirements and how can the effort be optimized?
- Flexibility and extensibility of approach: Is the approach scalable and how adaptable is it to changing requirements?
Recognizing and addressing early on the terminology challenges mentioned here – managing multiple terminologies, implementing standards, mapping local codes, and creating an implementation plan – will help organizations to reduce later complications and streamline their meaningful use implementation.
Susan Matney is a Medical Informaticist with 3M Health Information Systems.