Introduction and Objective
The HL7 Structured Product Labeling (SPL) standard1 implemented by the FDA uses the HL7 Reference Information Model (RIM)2 to represent the chemical and physical nature of medical products and their safe and effective use. While not all of this content is available today, we enrich the 3704 available SPLs with knowledge from the SPL terminology sources, including the VA’s NDF-RT3 and MeSH. To demonstrate the value of SPL for clinical decision support, we compared the performance of drug-intolerance (allergy) issue detection using SPL with the Regenstrief Institute (RI) own CPOE system, Gopher4 and its knowledge base.
Material and Methods
A de-identified dataset of intolerance records, medication order and supply records was extracted from the Regenstrief medical record system. 70% of the RI concepts in intolerance records and 76% of those in medication orders were mapped to MeSH. 60% of the National Drug Codes (NDC) in medication supply records (26% of distinct NDCs) could be found in one of the SPL labels. 76% of the Unique Ingredient Identifiers (UNII) used in SPL were mapped to MeSH. SPL labels enriched with MeSH annotations were loaded into an HL7 RIM-based database-system.5 With this database drug-intolerance issue detection can be done in SQL simply by joining the intolerance and the ordered drug through the transitive and reflexive IS-A relation of MeSH substance classes. The Gopher decisions support algorithms was emulated on the same database for comparison.
Results
The SPL/MeSH method detects twice the amount of patients with drug-intolerance issues and 4 times the number of issues than the current Gopher method that uses manually maintained drug sets. The effect is so strong that it overrides even the very incomplete mapping from RI dictionary terms and UNIIs to MeSH concepts. Even the unfortunate fact that less than 1/3 of distinct NDC codes were covered by current SPL content is outweighed by this effect. Most of the issues detected by SPL/MeSH were missed by Gopher due to the ad-hoc nature of its drug set definitions; especially multi-ingredient products were missing in the sets. The SPL method missed issues because of incomplete mapping of allergens, drugs, and UNIIs to MeSH.
Conclusion
Despite being severely inhibited by incomplete terminology mapping and coverage, a drug-intolerance detection method using SPL and its public terminology sources and implemented in a standard relational database finds several times more issues than an in-house maintained dictionary of ad-hoc drug sets. While most of the additionally detected issues are logically justified, the increased sensitivity highlights the importance of both well-maintained chemical structure terminology and more accurate intolerance records in order to increase the overall effectiveness of this safety feature.
Table 1.
Number of issues detected in the data counted by different Entities (Orders, Supplies, Patients, Drugs, and Allergens.)
| In … | Entities | this many Issues detected with | |||
|---|---|---|---|---|---|
| n | Gopher | SPL/MeSH | |||
| 2,734,787 | Orders | 10,239 | 0.4% | 45,129 | 1.7% |
| 51,143 | Patients | 4,368 | 9% | 8,832 | 17% |
| 1,469 | Drugs | 400 | 27% | 420 | 29% |
| 1,623 | Allergens | 270 | 23% | 375 | 23% |
| 3,682,926 | Supplies | 3,337 | 0.1% | 13,749 | 0.4% |
| 33,703 | Patient | 1,223 | 4% | 2,188 | 6% |
| 23,307 | Drugs | 659 | 3% | 464 | 2% |
| 1,623 | Allergens | 94 | 6% | 112 | 7% |
Acknowledgments
This work was performed at the Regenstrief Institute and is funded in part by the Agency for Healthcare Research and Quality (AHRQ) grant R01 HS15377 and the Food and Drug Administration (FDA). Without Randy Levin’s genial leadership this work would not have been possible.
References
- 1.Schadow G, Gitterman S, Boyer S, Dolin RH, editors. Ann Arbor, MI: Health Level Seven; 2005. HL7 v3.0 structured product labeling, release 3 [standard] [Google Scholar]
- 2.Schadow G, Russler DC, McDonald CJ. Conceptual Alignment of Electronic Health Record Data with Guidelines and Workflow Knowledge. Int J Med Inf. 2001 Dec;64(2–3):259–74. doi: 10.1016/s1386-5056(01)00196-4. [DOI] [PubMed] [Google Scholar]
- 3.Brown SH, Elkin PL, Rosenbloom ST, et al. VA National Drug File Reference Terminology: a cross-institutional content coverage study Medinfo 200411Pt 1477–81. [PubMed] [Google Scholar]
- 4.McDonald CJ, Tierney WM. The Medical Gopher – A Microcomputer System to Help Find, Organize and Decide About Patient Data. West J Med. 1986;145(6):823–829. [PMC free article] [PubMed] [Google Scholar]
- 5.Schadow G. Assessing the Impact of HL7/FDA Structured Product Label (SPL) Content for Medication Knowledge Management. Proc AMIA Symp. 2007 [PMC free article] [PubMed] [Google Scholar]