VNR code mapping to RxNorm

Details of VNR codes in finngen R6 mapped to standard RxNorm

The VNR codes are Nordic country-specific codes known as the Nordic Article Number. The VNR codes are 6-digit codes ranging from 000001-199999 and 370000-599999. They are assigned to all human medicines, veterinary medicines, herbal medicines, and traditional herbal medicines. Numbers outside this range are called ​​National Article Numbers which are used differently depending on the country.

RxNorm terminology on the other hand, is US specific terminology and provides normalized names for medications allows linking to many drug vocabularies commonly used in the US market.

We have mapped the Nordic country-specific VNR codes to RxNorm in FinnGen R6. Although the initial mapping was performed in R6 and is located in library-green in finngen_R6 folder the mapping can be used with any Data Freeze/Release.

The mapping and readme are located:

/finngen/library-green/finngen_R6/finngen_R6_medical_codes/fgVNR.tsv

/finngen/library-green/finngen_R6/finngen_R6_medical_codes/fgVNR_readme.txt

How the VNR code to RxNorm mapping was done:

1. VNR codes originating from different sources within FinnGen were combined into single table called 'OriginalVNR'

2. Additional information on VNR codes with missing drug name, strength, ingredient information were requested from Pharmaceutical Information Centre (Lääketietokeskus). This information is stored in the table 'ltklVNR'

3. Missing ingredient information of VNR codes was filled using ATC codes.

4. Administration routes, dosage forms and units were created for codes in both 'OriginalVNR' + 'ltklVNR' tables.

5. We processed source text format of Package, Substance, Substance strength, Administration Route and Dosage Form for both 'OriginalVNR' + 'ltklVNR' tables.

6. We used OHDSI drugmapping tool to map the parsed VNR code information to map to standard RxNorm

More details till step 5 from can be found in github repository.

Description of the columns of fgVNR.tsv are shown in below table.

Column Name	Column Type	Description	Example
VNR	INT64	six-digit VNR code.	518
ATC	STRING	ATC group code	N05AH04
MedicineName	STRING	Commerical Name	SEROQUEL
AdministrationRouteSourceTextFI	STRING	Administration route in text format as in source.	Suun kautta
AdministrationRoute	STRING	Valid value for administration route.	Oral use
DosageFormSourceTextFI	STRING	Dosage form in text format as in source.	tabletti, kalvopäällysteinen
DosageForm	STRING	Valid value for dosage form.	film-coated tablet
PackageSourceTextFI	STRING	Package info in text format as in source.	10 FOL
PackageSize	FLOAT64	Size of package in float format.	10
PackageFactor	INT64	Factor of package in float format.	1
PackageUnit	STRING	A valid unit value.	fol
SubstanceSourceTextFI	STRING	List of substances as in source.	quetiapine
Substance	STRING	Substance name. one row per substance.	quetiapine
SubstanceStrengthTextFI	STRING	Substance's strength in text format as in source.	25+100+200 mg
Strength	STRING	Mapped or fixed or split substance strength. If not then source strength used.	100 mg
SubstanceStrengthNumenatorValue	FLOAT64	Substance's strength value in numerator in float format.	100
SubstanceStrengthNumenatorUnit	STRING	A valid unit value.	mg
SubstanceStrengthDeominatorValue	FLOAT64	Substance's strength value in denominator in float format.	1
SubstanceStrengthDeominatorUnit	STRING	A valid unit value	1
ValidRange	BOOL	True if VNR is en the valid range (less than 200000 or between 370000 and 599999)).	TRUE
Source	STRING	From which table the code was taken.	ltklVNR or "originalVNR"
Status	STRING	How well the medicine has been processed	incomplete_dosageForm
VNRnew	STRING	A temporary VNR code created for drugs with single substance multiple strength values. Temporary VNR code will have letters a or b or c attached to the end.	000518a
calculateTotalStrength_message	STRING	How well the strength has been processed.	correct or "missmatch"
TotalStrength	FLOAT64	Total Strength of the drug which is PackageSize * PackageFactor * Dosage	10 * 1 * 100 = 1000
TotalStrengthUnit	STRING	Total Strength valid unit	mg
n_codes	INT64	Frequency of the VNR code	260
Dosage	FLOAT64	SubstanceStrengthNumenatorValue/SubstanceStrengthDeominatorValue	100/1 = 100
DosageUnit	STRING	A valid unit value	mg
MedicineNameFull	STRING	Commerical Name, Dosage Form and SubstanceStrength	SEROQUEL 25+100+200 mg

The fgVNR.tsv file was used as the input for OHDSI drugmapping tool. The tool requires VNR code with substance information followed by dosage form and drug strength. If no substance information is present then there will be no mapping.

• DrugMapping tool requires a Common Data Model (CDM) database with vocabulary data. To create the CMD, we:

  • Extracted CDM database schema from OHDSI common data model for Version 5.3.2 of CDM.

  • Created the CDM database schema in a PostgreSQL server Version 14.2

  • Changes were made in the CDM V5.3.2 SQL files generated from OHDSI common data model due to PostgreSQL server Version is > 9.
  • Downloaded the Vocabulary data of Default vocab list + Addition vocabularies for "Dosage Form" from Athena.
  • Uploaded the Vocabulary data from Athena to the PostgreSQL CDM v5.3.2 database
• Once the CDM database was up and running from PostgreSQL, we started setting up the DrugMapping Tool

  • DrugMapping Tool does following maps. We did "Clinical Drug Map"
  • Input file for DrugMapping tool was formatted to add missing columns
  • Information regarding the possible Clinical drug mapping possible along with total number of input drugs

    Input	Value
    Total Drugs	15,928
    Drugs with non-missing VNR Codes	15,902
    Unique VNR codes	14,655
    Drugs with non-missing VNR codes + Ingredient Codes	13,876
    Drugs with non-missing VNR codes + Ingredient Codes + dosage form	12,897
    Drugs with non-missing VNR codes + Ingredient Codes + dosage form + dosage value	12,644
    Drugs with non-missing VNR codes + Ingredient Codes + dosage form + dosage + dosage unit	12,638

• After fixing the Input file for DrugMapping Tool, it created three intermediary files

  • Ingredient Name Translation File

  • Unit Mapping File

  • Dose Form Mapping File

• All the three intermediary files need to be filled carefully

  • Ingredient Name Translation File - Simplest to fill using the input file

  • Unit Mapping File

    • Source units were to be mapped to standard units such as 'mg' and 'mL'. Example

      SourceUnit	DrugCount	RecordCount	Factor	TargetUnit	Comment
      %	303	1109084	0.01	mg/mg
      IU	309	400416	1	[U]
      U	25	360	1	[U]
      g	160	1614186	1000	mg
      g/l	8	1112	1	mg/mL
      mg	20905	68649900	1	mg
      mg/days	111	54182	0,289	mg/h
      mg/h	5	455	1	mg/h
      milli.IU	73	478234	1000000	[U]
      ml	5	4885	1	mL
      ug/puffs	6	212536	0.001	mg/{actuat}

    • Dose Form Mapping File

      • First thing is to extract all the dose form in domain "Drug" with concept_class "Dose Form" from all the vocabularies in the CDM database.

      • Second thing is to extract "relationship_id" of "Source - RxNorm eq" from CONCEPT_RELATIONSHIP table for all non-standard "Dose Form" from "additional vocabularies".

      • Match the cells in "DoseFrom" to the extracted standard dose forms which was only 49 out of 147 dose forms.

      • Manually filled out 85 dose forms with only 13 dose forms missing having low frequency. Example of filled dose form file can be seen below

        DoseForm	DrugCount	Priority	concept_id	concept_name	Comments
        BASIC CREAM	11		19082224	Topical Cream
        BATH ADDITIVE	1		19082228	Topical Solution
        BODY LOTION	1
        CAPSULE	279	0	19082168	Oral Capsule	Standard
        CAPSULE	279	1	19021887	Capsule	Non-Standard
        CAPSULE, HARD	664		19082168	Oral Capsule

    • The result of DrugMapping Tool after carefully filling out all three files can be shown below

      Percentage of possible drugs mapped is 12,089 of 12,638 (95.6%)

      Source drugs mapped to Clinical Drug	12089 of 14692 (82.283%)
      Source drugs mapped to Clinical Drug Form	562 of 14692 (3.825%)
      Source drugs mapped to Clinical Drug Comp	354 of 14692 (2.409%)
      Source drugs mapped to Ingredient	588 of 14692 (4.002%)
      Source drugs mapped Splitted	74 of 14692 (0.504%)
      Source drugs mapped Splitted Incomplete	3 of 14692 (0.02%)
      Source drugs mapped Total	13670 of 14692 (93.044%)
      Source drugs mapped to None	1022 of 14692 (6.956%)