=Paper=
{{Paper
|id=Vol-2931/ICBO_2019_paper_21
|storemode=property
|title=Comparing the Representation of Medicinal Products in RxNorm and SNOMED CT -
Consequences on interoperability
|pdfUrl=https://ceur-ws.org/Vol-2931/ICBO_2019_paper_21.pdf
|volume=Vol-2931
|authors=Jean Noel Nikiema,Olivier Bodenreider
|dblpUrl=https://dblp.org/rec/conf/icbo/NikiemaB19
}}
==Comparing the Representation of Medicinal Products in RxNorm and SNOMED CT -
Consequences on interoperability==
https://ceur-ws.org/Vol-2931/ICBO_2019_paper_21.pdf
Comparing the representation of medicinal products in RxNorm and SNOMED CT –
Consequences on interoperability
Jean Noel Nikiemaa, Olivier Bodenreiderb
a
Bordeaux Population Health Research Center, ERIAS, Univ. Bordeaux, Inserm UMR 1219, F-33000, Bordeaux, France
b
U.S. National Library of Medicine National Institutes of Health Bethesda, Maryland, USA
Abstract Interoperability among drug terminologies is especially important
for exchanging drug information internationally. For example, a
Objectives: To compare the representation of medicinal products
medication list established with RxNorm in the U.S. could be
in RxNorm and SNOMED CT and assess the consequences on in-
made available to any electronic health record (EHR) system in
teroperability. Methods: To compare the two models, we manu-
the world, in which drugs are represented using SNOMED CT.
ally establish equivalences between the types and definitional fea-
To fully support this use case, however, the models of medicinal
tures of medicinal products entities in RxNorm and SNOMED CT.
products in RxNorm and SNOMED CT must be compatible, such
We highlight their similarities and differences. Results: Both
that one can be accurately translated into the other.
models share major definitional features including ingredient (or
substance), strength and dose form. SNOMED CT is more rigor- We focus on RxNorm and SNOMED CT, because RxNorm is the
ous and better aligned with international standards. In contrast, standard drug terminology in the U.S. and SNOMED CT is the
RxNorm contains implicit knowledge, simplifications and ambi- largest clinical terminology in the world, supported by a consor-
guities, but its model is simpler. Conclusions: Since their models tium of over 40 countries. While the RxNorm model has been an-
are largely compatible, medicinal products from RxNorm and alyzed (5,6), and reused to create others standards (7,8) and to
SNOMED CT are expected to be interoperable. However, specific integrate drug terminologies worldwide (8), there has not been a
aspects of the alignment between the two models require particu- detailed comparison between RxNorm and SNOMED CT. More-
lar attention. over, the SNOMED CT model for medicinal products is particu-
larly interesting, because it was recently updated, in part to com-
Keywords:
ply with IDMP requirements (9).
RxNorm; SNOMED CT; medicinal products.
In this investigation, we compare the representation of medicinal
products in RxNorm and SNOMED CT. The objective of our
Background work is to analyze their similarities and differences and the con-
sequences of these differences on interoperability between the
Drug terminologies, such as RxNorm and the medicinal product two terminologies.
hierarchy of SNOMED CT (Systematized Nomenclature of Med-
icine-Clinical Terms), support multiple use cases, including elec- Methods and results
tronic prescription, drug information exchange, medication rec-
onciliation, and analytics (including pharmacovigilance) (1,2). A
In this section, we describe the models of RxNorm and SNOMED
formal representation of medicinal products is needed for the
CT with focus on their definitional characteristics. Then we iden-
principled development and maintenance of such drug terminolo-
tify similarities and differences between the two models.
gies, as well as for precisely aligning existing drug terminologies
(3). The SNOMED CT model for medicinal products
Many definitional characteristics of medicinal products are simi- The SNOMED CT, the largest clinical terminology in the world,
lar among drug terminologies. For example, clinical drugs are is an international clinical terminology based on a formal concept
generally defined in terms of ingredient, strength and dose form. model (10). SNOMED CT recently published a new model for the
However, the level of formality and the formalism used for rep- representation of medicinal products integrating requirements
resenting medicinal products may differ among terminologies. from IDMP (9). The model was developed to support interna-
Some attributes may also be specific to some terminologies (es- tional usage. Therefore, it is restricted to generic drugs and does
pecially for country-dependent attributes, such as packaging in- not represent packaging information or branded drugs, which tend
formation). to be country-specific.
In addition to existing drug terminologies, international standards In accordance with requirements from IDMP, clinical drugs are
have been developed for the representation of medicinal products, represented in a closed worldview. This means that characteristics
such as IDMP (Identification of Medicinal Products). IDMP (4), used to define clinical drugs must be sufficient and what is not
a collection of recommendations from the International Standards stated is false. In contrast, in the open worldview, what is not
Organization (ISO).
Copyright © 2019 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
stated is potentially true. For example, the representation of a clin- x
One clinical drug entity, in closed worldview only
ical drug containing Atorvastatin must clearly state that this prod- (e.g., 320818006 | Product containing precisely cetiriz-
uct only contains the substance Atorvastatin as its active ingredi- ine hydrochloride 10 milligram/1 each conventional re-
ent (i.e., without any other active ingredient). In the open lease oral tablet (clinical drug)).
worldview, products containing Atorvastatin could also contain The representation of SNOMED CT entities is based on "defini-
other active ingredients, e.g., Amlodipine. tional roles" and related "types of values" in SNOMED CT (Fig-
As shown in Figure 1, the representation of medicinal products in ure 1):
SNOMED CT is based on a model with six (6) entities, arranged x Substance is the type of values for the active ingredi-
in a subclass hierarchy: ent, precise active ingredient and basis of strength
x Two medicinal product entities, in open and closed roles, for example 372523007 | Cetirizine (substance)
worldview (e.g., open worldview: 108655000 | Product and 108656004 | Cetirizine hydrochloride (substance).
containing cetirizine (medicinal product) and closed (The basis of strength is the substance in reference to
worldview: 775140005 | Product containing only which strength is defined.)
cetirizine (medicinal product)). x Unit of measure is the type of values for the strength
x Two medicinal product form entities, in open and unit roles, for example, 258684004 | milligram (quali-
closed worldview, (e.g., open worldview: 768065006 | fier value).
Product containing cetirizine in oral dose form (medic- x Number is the type of values for the strength value
inal product form) and closed worldview: 778701007 | roles, for example, 3445001 | 10 (qualifier value).
Product containing only cetirizine in oral dose form
(medicinal product form)). x Pharmaceutical dose form is the type of values for the
manufactured dose form role, for example, 421026006 |
x One medicinal product precisely entity in closed Conventional release oral tablet (dose form).
worldview only (optional entity, currently not repre-
sented in SNOMED CT – hypothetical example: x Unit of presentation is the type of values for the unit
Product containing only cetirizine hydrochloride of presentation role, for example, 732936001 | Tablet
(medicinal product)). (unit of presentation).
Open-world entities Closed-world entities Definitional features
Medicinal product
(MP)
MP/MPO:
108655000 | Product • Active ingredient (substance)
containing cetirizine Medicinal product
(medicinal product) |
Only (MPO)
775140005 | Product
containing only cetirizine
(medicinal product) |
Medicinal product MPP:
precise only (MPP) • Precise active ingredient (substance)
(not represented) | Product
containing only cetirizine
Medicinal product hydrochloride (medicinal
product) |
Form (MPF) MPF/MPFO:
• Active ingredient (substance)
768065006 | Product • Dose form (pharmaceutical dose form)
containing cetirizine in oral
dose form (medicinal product Medicinal product
form) |
form only (MPFO) CD:
778701007 | Product • Precise Active ingredient (substance)
containing only cetirizine in • Dose form (pharmaceutical dose form)
oral dose form (medicinal • Basis of strength (substance)
product form) | Clinical drug
• Strength units (unit of measure)
IS-A • Strength values (numbers)
320818006 | Product containing • Units of presentation (units of presentation)
precisely cetirizine hydrochloride
10 milligram/1 each conventional
release oral tablet (clinical drug) |
Figure 1– SNOMED CT model for the representation of medicinal products showing the six types of entities defined in the model,
along with their definitional features and examples from the SNOMED CT terminology
Examples RxNorm generic drug entities Definitional features
Cetirizine IN:
[RxCUI = 20610] Ingredient • Ingredient
has_ingredient
SCDC:
cetirizine hydrochloride 10
MG [RxCUI = 1011480] Clinical Drug Component has_ingredient • Ingredient
• Strength
SCDF:
Cetirizine Oral Tablet
[RxCUI = 371364] consists_of Clinical Drug Form • Ingredient
• Dose form
isa SCD:
• Ingredient
cetirizine hydrochloride 10
• Strength
MG Oral Tablet Clinical Drug • Dose form
[RxCUI = 1014678]
• Quantity factor (optional)
• Qualitative distinction (optional)
Figure 2– Simplified RxNorm model for the representation of generic medicinal products showing the four types of entities defined in
the model, along with their definitional features and examples from the RxNorm terminology
Closed-worldview are “closed” with respect to their active ingre- Cetirizine [RxCUI = 20610], PIN: cetirizine hydrochlo-
dient(s). More specifically, medicinal product and medicinal ride [RxCUI = 203150], MIN: Cetirizine /
product form entities are closed with respect to their active ingre- Pseudoephedrine [RxCUI = 352367])
dient(s), while medicinal product precisely and clinical drug enti- x Clinical drugs component (SCDC), combining
ties are closed with respect to their precise active ingredient(s). ingredient and strength (e.g., cetirizine hydrochloride
There are no hierarchical relations among substances. However, 10 MG [RxCUI = 1011480])
there is a "modification of" relation between a modified substance x Clinical drugs form (SCDF), combining ingredient and
(e.g., ester or salt) and the corresponding base substance (e.g., be- dose form (e.g., Cetirizine Oral Tablet [RxCUI =
tween Atorvastatin calcium and Atorvastatin). Modified sub- 371364])
stances can be further modified.
x Clinical drug (SCD), combining ingredient, strength
IDMP requires that dose forms be defined in reference to a list of
and dose form (e.g., cetirizine hydrochloride 10 MG
dose forms from the European Directorate for Quality in Medi-
Oral Tablet [RxCUI = 1014678])
cines (EDQM). EDQM distinguishes between dose forms and
units of presentation. Units of presentation are used to express the The representation of these entities relies on three mandatory and
strength and quantity in countable entities, while dose forms cor- two optional definitional features:
respond to the physical structure of the medicinal product. x Mandatory definitional features:
In accordance with requirements from IDMP, strength units in ingredient (IN/PIN/MIN) (e.g., IN: Cetirizine
SNOMED CT are aligned with the international standard for units [RxCUI = 20610], PIN: cetirizine hydrochloride
of measure, UCUM (Unified Code for Units of Measure). [RxCUI = 203150], MIN: Cetirizine /
Finally, depending on the unit of presentation, strength can be Pseudoephedrine [RxCUI = 352367])
represented as concentration strength, presentation strength or dose form (DF) (e.g., Oral Tablet [RxCUI =
both. 317541])
The RxNorm model strength (e.g., 10 MG)
Created in 1992, RxNorm is a normalized terminology for clinical x Optional definitional features (see below for examples):
drugs in the U.S. RxNorm represents both generic drugs and quantity factor (QF)
branded drugs, as well as packs (11). The full model of RxNorm qualitative distinction (QD)
contains ten entities, five for generic drug entities and five for
branded drugs entities. For comparison with SNOMED CT, we Strength in RxNorm is normalized. In its units of measure (e.g.,
only present RxNorm generic drug entities and also omit packs. for volume, weight, surface), RxNorm uses one unit for each type
quantity (e.g., milligram for weight rather than gram or mi-
The simplified RxNorm model for generic drug entities includes crogram).
four entities (Figure 2):
The representation of dose forms in RxNorm is not based on a
x Ingredient, including base ingredient (IN), precise specific standard (12). It is also important to note that the SCDs
ingredient (PIN), and multi-ingredient (MIN) (e.g., IN:
Definitional features
RxNorm entities SNOMED CT entities
SNOMED CT RxNorm
IN/PIN Ingredient (IN)/ IN/PIN (Medicinal
Substance Medicinal product
(Substance) Precise ingredient (PIN) product)
Medicinal product
Pharmaceutical only
Dose Form
Medicinal product
Dose Form (DF) Clinical drug Form (SCDF)
form
Unit of
presentation
Medicinal product
form only
Unit of
Strength Unit
measure
Clinical drug (SCD) Clinical drug
Number Strength Value
Correspondence between Definitional Correspondence
Legend
definitional features feature between entities
Figure 3– Correspondence between the RxNorm and SNOMED CT models
and SCDCs refer to the basis of strength substance (e.g., cetirizine Comparison of the RxNorm and SNOMED CT models
hydrochloride), while SCDFs refer to the base ingredient (e.g.,
cetirizine). Of note, ingredients in RxNorm can (purposely) be To compare the two models, we manually establish equivalences
understood as either the substance contained in a medicinal prod- between their entities and between their definitional features,
uct as active ingredient (e.g., “cetirizine the substance”) or the based on our analysis of the two models.
class of all medicinal products containing this substance as active First, we need to disambiguate the notion of ingredient in
ingredient. Precise ingredients (PINs) generally correspond to RxNorm (IN,PIN, MIN), because, as mentioned earlier, it can be
modified forms of the corresponding base ingredients (INs). PINs understood as either a substance or a class of medicinal products.
cannot be further modified. Therefore, as shown in Figure 3, ingredients in RxNorm corre-
In addition, RxNorm does not explicitly have a notion of spond to SNOMED CT medicinal products (in open and closed
"worldview" (i.e., open or closed worldview) for its entities. worldview) or to SNOMED CT substances, which are active in-
While clinical drugs implicitly refer to a closed worldview, ingre- gredients of SNOMED CT medicinal products. In practice,
dients, clinical drug components and clinical drug forms can be RxNorm ingredients are often associated with multiple SNOMED
understood in both open and closed worldview, leaving it to que- CT entities, typically with one substance entity and one medicinal
ries to distinguish between the two. product entity. Disambiguation consists in identifying which
SNOMED CT entity comes from the substance hierarchy (and
Finally, the Quantity Factor (QF) is a number followed by a unit
treating it as a value for the definitional feature “active ingredi-
of measure corresponding to vial sizes or patch durations (e.g.,
ent”), while the SNOMED CT entity corresponding to an entity
"12H"). RxNorm does not explicitly state whether strength is ex-
from the medicinal product hierarchy is marked as an asserted
pressed as presentation strength or concentration strength. Presen-
equivalence for the RxNorm medicinal product entity.
tation strength can be derived from concentration strength by mul-
tiplying the concentration strength by the quantity factor. (For ex- RxNorm does not formally have the notion of "unit of presenta-
ample, if the concentration strength is 1MG/ML and the QF is tion". Units of presentation are implicitly represented through
2ML, the presentation strength is 2MG/2ML). The Qualitative dose forms in RxNorm, whereas the two notions are represented
Distinction (QD) corresponds to some qualitative characteristic of separately in SNOMED CT. For example, in SNOMED CT, tab-
a drug outside the main definitional features (e.g., "sugar free" let is the logical "unit of presentation" of the conventional release
and “abuse-deterrent”). QD and QF are optional modifiers used oral tablet, while the two are conflated in the RxNorm dose form
in RxNorm to define medicinal products when it is clinically rel- “Oral Tablet”. Therefore, RxNorm dose forms generally corre-
evant to identify such distinctions (12). spond to pairs of a pharmaceutical dose form and a unit of presen-
tation in SNOMED CT.
In addition, there are no materialized entities for SCDCs in
SNOMED CT. Instead, strength and basis of strength substance
are associated as part of the definition of a clinical drug in
SNOMED CT. Therefore, SCDCs cannot be related to entities in Strength entities require minimal attention, specifically for con-
SNOMED CT, but their defining features are represented as part verting RxNorm “fixed unit” into the clinically appropriate unit
of clinical drug entities. used in SNOMED CT. Simple arithmetic is also required to con-
SCDs in RxNorm are equivalent to clinical drugs in SNOMED vert concentration strength and quantity factor in RxNorm to
CT as they essentially share the same definitional features. The presentation strength in SNOMED CT wherever appropriate.
quantity factor in RxNorm has no direct equivalent in SNOMED In contrast, aligning dose forms requires more analysis, as
CT, but QF information is implicitly represented in the presenta- RxNorm dose forms generally correspond to pairs of a pharma-
tion strength. In contrast, qualitative distinctions are absent from ceutical dose form and a unit of presentation in SNOMED CT.
the SNOMED CT model. The absence of correspondence for qualitative distinction in
While RxNorm only represents one level of modification (be- SNOMED CT may lead to multiple clinical drugs in RxNorm
tween PIN and IN), SNOMED CT can represent arbitrary levels mapping to a single clinical drug in SNOMED CT. For example,
of modification among substances. the distinction between Cholestyramine Resin 4000 MG Powder
Both RxNorm and SNOMED CT have the notion of concentration for Oral Suspension [RxCUI = 848943] and its sugar-free form
strength and presentation strength. However, RxNorm empha- Sugar-Free Cholestyramine Resin 4000 MG Powder for Oral
sizes concentration strength (from which presentation strength Suspension [RxCUI = 1801279] in RxNorm is lost in SNOMED
can be calculated using the quantity factor), whereas SNOMED CT. This issue is unlikely to result in clinically significant align-
CT explicitly represent both presentation strength and concentra- ment errors.
tion strength when necessary. The absence of materialization of the clinical drug component
Finally, RxNorm normalizes all quantities to one unit (per type of (SCDC) entity in SNOMED CT does not create an alignment is-
quantity), whereas SNOMED CT uses units that are most clini- sue, because SCDCs are essentially navigational entities in
cally appropriate (following IDMP requirements). For example, RxNorm. They are not crucial to any of the main use cases for
RxNorm uses 0.001 milligram and SNOMED CT 1 microgram. RxNorm or SNOMED CT.
This difference merely reflects differences in editorial guidelines, Future work. In future work, we plan to translate RxNorm into
as conversion between the two is trivial. the SNOMED CT model for medicinal products. The resulting
alignment would make RxNorm entities directly compatible with
Discussion SNOMED CT’s. One benefit of this alignment would be to assess
interoperability between RxNorm and SNOMED CT, potentially
Findings. Not surprisingly, the models used by RxNorm and enriching SNOMED CT with clinical drugs currently specific to
SNOMED CT for representing medicinal products are fairly sim- RxNorm. Additionally, this alignment would offer an opportunity
ilar and essentially compatible. Both models share major defini- for quality assurance by identifying cases where alignment is ex-
tional features including ingredient (or substance), strength and pected, but cannot be inferred (e.g., because of a difference in ba-
dose form. Only the qualitative distinction feature of RxNorm has sis of strength substance for a given clinical drug between
no correspondence at all in SNOMED CT. RxNorm and SNOMED CT).
SNOMED CT is more rigorous and better aligned with interna-
tional standards. In SNOMED CT, differences tend to be made Conclusion
explicit, e.g., between a substance and the class of medicinal
products containing this substance as an ingredient, or between In this investigation, we examined the similarities and differences
the class of all medicinal products containing only a given active between the representation of medicinal products in RxNorm and
ingredient and the class of all medicinal products containing at SNOMED CT. We established that both models share major def-
least this active ingredient . SNOMED CT also offers more flex- initional features including ingredient (or substance), strength and
ibility with relations among substances, as opposed to a fixed pre- dose form. Because of subtle differences between the two models,
cise ingredient to base ingredient relationship in RxNorm. This specific aspects of their alignment require particular attention.
precision comes at the price of a more complex model, and pos-
sibly a steeper learning curve. In contrast, RxNorm contains im- Acknowledgment
plicit knowledge, simplifications and ambiguities, but its model
is simpler. This work was supported by the Intramural Research Program of
With features, such as explicit closed worldview for clinical drug the NIH, National Library of Medicine. The authors would like to
entities, use of standard dose forms from EDQM, use of UCUM thank the developers of RxNorm and the contributors to the
units, and use of clinically appropriate strength values, SNOMED SNOMED International Drug Model Working Group for useful
CT shows better compliance with international standards (namely discussion We are particularly grateful to Tammy Powell and
IDMP) than RxNorm does. Chris Hui (RxNorm), and Julie James, Jim Case, Yongsheng Gao,
Emma Melhuish, Toni Morrison, Guillermo Reynoso, Farzaneh
Consequences on alignment. Since their models are largely com-
Ashrafi and Phuong Skovgaard (SNOMED CT). We also thank
patible, medicinal products from RxNorm and SNOMED CT are
NLM colleagues, Lee Peters, Robert Wynne and Phill Wolf, for
expected to be interoperable. However, specific aspects of the
useful discussion.
alignment between the two models require particular attention.
The values of ingredient can be aligned rather trivially (after dis-
ambiguation between the two meanings of RxNorm ingredients, Address for correspondence
substance and class of medicinal products containing this sub-
stance as an ingredient). Jean.nikiema@u-bordeaux.fr
Olivier.bodenreider@nih.gov
References 6. Liu S, Wei Ma, Moore R, Ganesan V, Nelson S. RxNorm:
prescription for electronic drug information exchange. IT
1. Lupse O-S, Chirila C-B, Stoicu-tivadar L. Harnessing On- Professional. 2005 Sep;7(5):17–23.
tologies to Improve Prescription in Pediatric Medicine. 7. Wang L, Zhang Y, Jiang M, Wang J, Dong J, Liu Y, et al.
Studies in Health Technology and Informatics. 2018;97– Toward a normalized clinical drug knowledge base in
101. China—applying the RxNorm model to Chinese clinical
2. Farrish S, Grando A. Ontological approach to reduce com- drugs. Journal of the American Medical Informatics Associ-
plexity in polypharmacy. AMIA Annu Symp Proc. ation. 2018 Jul 1;25(7):809–18.
2013;2013:398–407. 8. Hanna J, Joseph E, Brochhausen M, Hogan WR. Building a
3. Lai EC-C, Ryan P, Zhang Y, Schuemie M, Hardy NC, Ka- drug ontology based on RxNorm and other sources. Journal
mijima Y, et al. Applying a common data model to Asian of Biomedical Semantics. 2013;4(1):44.
databases for multinational pharmacoepidemiologic studies: 9. Bodenreider O, James J. The New SNOMED CT Interna-
opportunities and challenges. Clinical Epidemiology. 2018 tional Medicinal Product Model. In: Proceedings of the In-
Jul;Volume 10:875–85. ternational Conference on Biological Ontology (ICBO
4. European Medicines Agency. Introduction to ISO Identifi- 2018). Oregon, USA,; 2018.
cation of Medicinal Products, SPOR programme [Internet]. 10. Héja G, Surján G, Varga P. Ontological analysis of
2016. Available from: https://www.ema.europa.eu/docu- SNOMED CT. BMC Medical Informatics and Decision
ments/other/introduction-iso-identification-medicinal-prod- Making. 2008;8(Suppl 1):S8.
ucts-spor-programme_en.pdf 11. Nelson SJ, Zeng K, Kilbourne J, Powell T, Moore R. Nor-
5. Dhavle AA, Ward-Charlerie S, Rupp MT, Kilbourne J, malized names for clinical drugs: RxNorm at 6 years. Jour-
Amin VP, Ruiz J. Evaluating the implementation of nal of the American Medical Informatics Association. 2011
RxNorm in ambulatory electronic prescriptions. Journal of Jul;18(4):441–8.
the American Medical Informatics Association. 2016 12. Bodenreider O, Cornet R, Vreeman D. Recent Develop-
Apr;23(e1):e99–107. ments in Clinical Terminologies — SNOMED CT, LOINC,
and RxNorm. Yearbook of Medical Informatics. 2018
Aug;27(01):129–39.