Maintenance and Update Issues for Clinical Vocabulary Mapping

Patricia S. Wilson, RT (R), CPC, PMP
3M Health Information Systems

Presentation for the Clinical Vocabulary Mapping Methods Symposium
AHIMA Conference, 2005
San Diego, California
October 15, 2005

Introduction

The Electronic Health Record (EHR) is no longer an option. It has become a necessity. It is essential to an enterprise to have interoperability between existing medical systems within the enterprise as well as with medical and billing organizations in the community. Many of these enterprises have realized that mapping is required in order to assist with the goal of semantic interoperability. Mapping is the process of creating one-way links between concepts and terms for specific purposes, often involving patient, administrative or interface contexts. The process of mapping can be:

1. Between concepts in a reference terminology and external administrative or reimbursement classifications,
2. Between concepts in a reference terminology and legacy terms/codes
3. Between concepts in different source terminologies or application vocabularies

The one-way links can be through representation synonymy, term association, relationship, attribute, layers of granularity, composition/decomposition, etc. The result is a universal cross-reference map accounting for all concepts and terms. An integral part of the mapping process is the maintenance and updates of mapped data. It is essential that the integrity of the mapped data be preserved while providing a timely and complete update of new, 'deleted' and/or changed terms and relationships.

Discussion

The process of mapping is dynamic. Rarely does a reference terminology, classification system or legacy code set remain static. The terms are updated on a schedule established by the organization that manages the data set. For example, Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT®) is updated semi-annually. Common Procedural Terminology (CPT®) is updated quarterly. Logical Observation Identifiers Names and Codes (LOINC®) is updated as needed, approximately 3 times a year. Since mapping involves a source and a target, the ability to maintain a map between SNOMED CT and LOINC, for example, would require a synchronization of the update schedules for the 2 data sets. The complexity of maintenance grows exponentially depending on how many source-target maps have been undertaken by an enterprise.

In order to assure the integrity of the mapped data while staying current with the updated information for the source and target, quality assurance protocols, a mechanism for logging status of work, and well-organized version control are required.

Quality Assurance: The basis for mapping quality assurance is the heuristics, use case, data structure and database protocol for the intended map. This basic information should be evaluated for any risks that could arise if all of the rules and assumptions are not properly or completely adhered to. A set of sequenced test data should be built based upon the identified risks. The test data should be run on a routine basis against the mapped data at any time that a possible change to the map may have occurred.

A companion to the test suite is a history log of all transactions made against the data map. This log should consist of all possible transactions that could occur for any single item within the data structure of the map. The log should include an identifier of all persons who have access to the data map, when the data was accessed and what transpired. Security protocol and backup procedures are also essential to a good quality assurance process.

Logging of work: Maintaining a log of work assignments is important for making sure that all mapping is performed. This may be a straightforward process if the number of persons involved in the mapping process is minimal (less that 5). As the staff increases, so do the complexity of workflow management. The method by which the workload is divided, performed and checked will require a work log. If the number of staff is small, the work may be broken down based on specialty or preference of the mappers and could be handled manually through a 'check-in/check-out' library or file. The ideal for a larger staff is to have an automated mechanism for receipt, formatting, dispensing, and check-off of completed work. This will also assist in monitoring of mappers who may require assistance.

Version Control: The process of version control involves not only the receipt of data but also the dissemination of the mapped data to the end-user upon completion of mapping. Since the enterprise performing the map cannot control the frequency and amount of new, changed, and/or deleted data that will impact the source or target data-set, mappers need to be prepared for any amount of data change that might occur. The receipt of data should be directly tied to the logging of work. It should assure that the workflow is consistent with the availability of the mapping staff. Part of the work log should be the schedule of the source and target data updates. This should include the method of receipt of data, a contact person, the structure in which the data will be received, and whether or not the data requires a license for use.

If the data is being received from within the enterprise, a schedule protocol should be implemented for each legacy system. A protocol should assure that the legacy system does not begin using data until it is mapped in order to prevent undue errors in the sending and storage process of the EHR. The schedule should allow for a buffer of time from the receipt of the data for mapping until the data can be made available for use by the department. Legacy data will often require a means for sending questions to obtain clarification of the data sent. The mappers should be familiar with the legacy system and have a point of contact at the department to assist with any problems that may arise.

A timely and efficient release of updated maps to the end user is directly tied to the timely receipt of updated data sets. It is possible that many systems within the enterprise will come to depend on the mapping to provide updates for billing and regulatory data sets. Mapping updates will need to be updated, tested, and ready for release with enough time for the departments to download and test the data for their own use prior to any imposed deadlines. This will require a finely tuned process of monitoring over the entire maintenance and updating process.

Conclusion

The process of mapping has become an essential part of an enterprise's ability to achieve interoperability. The creation, maintenance, and updating of these one-way links between a source and target data is a dynamic process. It is imperative that the integrity of the mapped data remains intact in order to preserve the clinical and administrative viability of the data that may be stored using the maps. Lack of control over the maintenance and updating of mapped data can result in incorrect data being stored in a patient's record. Good maintenance and update procedures and practices will assure that the integrity of the mapping remains intact.

References

Shaun C. Shakib, Elva Knight, Joni Endo, Lee Min Lau. An Application to Integrate the Logistical and Technical Aspects of Data Dictionary Support to Multiple Healthcare Systems. Journal of the American Medical Informatics Association, Proceedings of the 2002 Annual Symposium.

Kasey B. Poon, Shaun C. Shakib, Lee Min Lau. A Quality Assurance Browser Tool for Tracking Vocabulary Mapping. Journal of the American Medical Informatics Association, Proceedings of the 2004 Annual Symposium.

Shaun C. Shakib, Kasey B Poon, Lee Min Lau. Tools and Processes to Improve Data Mapping Accuracy and Reliability. Journal of the American Medical Informatics Association, Proceedings of the 2004 Annual Symposium.

Kasey B. Poon, Chengjian Che, Kent Monson, Shaun C. Shakib, Lee Min Lau. The Evolution of Tools and Processes for Data Mapping. To be published in the Journal of the American Medical Informatics Association, Proceedings of the 2005 Annual Symposium.

Chengjian Che, Kent Monson, Kasey B. Poon, Shaun C. Shakib, Lee Min Lau. Managing Vocabulary Mapping Services. To be published in the Journal of the American Medical Informatics Association, Proceedings of the 2005 Annual Symposium.

© 3M HIS