• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of procamiaLink to Publisher's site
AMIA Annu Symp Proc. 2006; 2006: 1159.
PMCID: PMC1839288

Mapping Concepts in Medical Error Taxonomies

Min Zhu, MD, MS,1 Yong Huang, MD,2 Yang Gong, MD, MS,1 Juliana J Brixey, MPH, MSN, RN,1 Jiajie Zhang, PhD, and James P. Turley, PhD, RN1

Background

Developing a domain ontology is a complex task. It is necessary to clarify domain concepts and the relationships between and among the concepts. Several taxonomies exit for the medical error domain. However, these taxonomies are divergent in granularity and asymmetry (assigning terms under different categories in different taxonomies). It is difficult to map concepts in different taxonomies to each other [1, 2]. Our research team selected the following listed eight taxonomies in published literature to merge into a medical error ontology.

  1. The JCAHO patient safety event taxonomy
  2. NCC MERP Taxonomy of Medication Errors
  3. Taxonomy of Nursing Errors (TNE)
  4. A Preliminary Taxonomy of medical errors in Family Practice3 (PTFP)
  5. Cognitive Taxonomy of Medical Errors23 (COG)
  6. Taxonomy of Medical Errors for Neonatal Intensive Care24 (NIC)
  7. Australian Patient Safety Foundation Taxonomy (APSF)
  8. Pediatric Patient Safety Taxonomy (PED)

In this poster, we report on the solutions to the problems of granularity and asymmetry found in these taxonomies when mapping concepts within the ontology.

Method

The goal is to keep the mapping process complete and consistent. We started NCC MERP and paired its concepts with the other 7 taxonomies one by one. We then selected the JACHO from the 7 taxonomies and paired its concepts with the rest six taxonomies one by one. We continued this process until all taxonomies had been paired with each other. We put all concepts of the source taxonomies into Protégé as classes. We examined each paired taxonomies and selected relevant classes. We found the relevant classes were mostly in the sections of Error Type, Location-Service, and Patient Outcome.

We classified relevant classes as perfect matches, unmatches with granularity issue and/or asymmetry issue. If the definition of the source class was the same as the definition of target class, we judged these two classes were perfect match and used the property isEquivalentTo to link them. For example, ‘23.7.18, Sub-acute Care’ in NCC MERP is equivalent to ‘3.01.01.04 Subacute Care’ in JACHO.

We took the pair NIC and NCC MERP as an example of the granularity problem in the taxonomies. NIC represents ‘(2) 1 Neonatal intensive care unit’ and ‘(2) 2 Intermediate care or step-down unit’ as two classes, while NCC MERPP used ‘23.7.4.3 Neonatal ICU/Step Down (Infant Transitional)’ as one class.

After discussion, we addressed granularity issue by one-way mapping. We created a new property isCoveredBy to relate more granular items to less granular ones, this was a unidirectional relationship. ‘(2) 1 Neonatal intensive care unit’ and ‘(2) 2 Intermediate care or step-down unit’ in NIC were separately covered by ‘23.7.4.3 Neonatal ICU/Step Down (Infant Transitional)’ in NCC MERP.

An asymmetry example occurs where ‘70 Type’ which means error type in NCCMERP involved 14 sub categories, while ‘2 type’ in JACHO involved 3 sub categories and ASPF only had a simple class ‘1.1 Medication’.

To solve the asymmetry issue, we used ‘[exists](some value isCoveredBy (Property) Type(Class Name)’ to link relevant classes as asserted conditions which means some value in the current class can be covered by the linked class.

Conclusion

This approach to mapping works linked the eight source taxonomies together with our ontology for better interoperability.

Future work

We are collecting medical error cases to test the ontology development work. Some cases have been coded in one of these taxonomies. We will then use our medical error ontology as framework to integrate these cases and evaluate the precision and accuracy of this mapping work. When the ontology is accomplished, we will develop a medical error reporting system based on our ontology to collect, organize and analyze medical error data.

Acknowledgement

This project was supported by NLM Grant R01 LM007894-01A1

References

1. Brixey J, Johnson TR, Zhang J. Evaluating a medical error taxonomy. Proc AMIA Symp. 2002:71–5. [PMC free article] [PubMed]
2. Boxwala AA, Dierks M, Keenan M, et al. Organization and representation of patient safety data: current status and issues around generalizability and scalability. J Am Med Inform Assoc. 2004 Nov–Dec;11(6):468–478. [PMC free article] [PubMed]

Articles from AMIA Annual Symposium Proceedings are provided here courtesy of American Medical Informatics Association
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Links

  • PubMed
    PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...