Build a list of antibiotics code in RxNorm

[razekmh]

The watch lists of antibiotics are all in ATC which is not really useful when searching the drug_exposure table. At the moment I would take an ATC code and search in the athena database or use the drug_lookup_table neither is an effective method. I should find a way to pull the codes without having to search one by one like this.

[razekmh]

A bit of a background. I have a list of the antibiotics on the UK watchlist. The list is of the names of the drugs and ATC codes. I believe the list comes from a paper on the subject.

I want to search the drug_exposure table to find any instances of the antibiotics in the data we have access to. However, it is not possible to search directly using the ATC codes. I have to find the RxNorm codes for each of these drugs. I have two obvious ways to do this.

use the omop_drug_lookup_create

I can filter the table using the ATC codes to find equivalent RxNorm codes. I then used the extracted RxNorm codes to filter the drug_exposure table. Alas, No matches!

> drugclasses <- omopcept::omop_drug_lookup_create()
> lookup_uk_watchlist <- filter(drugclasses, ATC_code %in% UK_ATC_watchlist$`ATC code`)
> filter(drug_exposure_df, drug_exposure_df$drug_concept_id %in% lookup_uk_watchlist$drug_concept_id)
# A tibble: 0 × 25

use athena

This offers more codes to look for. For example looking at J01CA04 for amoxicillin

• omop_drug_lookup_create gets me one match

> filter(drugclasses, ATC_code == "J01CA04")
# A tibble: 1 × 7
  drug_concept_name drug_concept_id
  <chr>                       <int>
1 amoxicillin               1713332
# ℹ 5 more variables: drug_concept_class_id <chr>,
#   ATC_level <chr>, ATC_concept_name <chr>,
#   ATC_code <chr>, ATC_concept_id <int>

Which naturally returns null

> filter(drug_exposure_df, drug_exposure_df$drug_concept_id == "1713332")
# A tibble: 0 × 25
# ℹ 25 variables: drug_exposure_id <int>,
#   person_id <int>, drug_concept_id <int>,
#   drug_exposure_start_date <date>,
#   drug_exposure_start_datetime <dttm>,
#   drug_exposure_end_date <date>,
#   drug_exposure_end_datetime <dttm>,
#   verbatim_end_date <date>, …

however checking in athena I can get a list of around 441 matches most of these are commercial names for drugs with amoxicillin in them for example Amoxicillin / Clarithromycin / Omeprazole Oral Tablet, Amoxicillin / Clavulanate Intravenous Solution, Amoxicillin / Clavulanate Powder for Oral Suspension .. I copied all to a temp table and still found nothing

> filter(drug_exposure_df, drug_exposure_df$drug_concept_id %in% tmp_amx_data$Concept.id)
# A tibble: 0 × 25
# ℹ 25 variables: drug_exposure_id <int>,
#   person_id <int>, drug_concept_id <int>,
#   drug_exposure_start_date <date>,
#   drug_exposure_start_datetime <dttm>,
#   drug_exposure_end_date <date>,
#   drug_exposure_end_datetime <dttm>,
#   verbatim_end_date <date>, …

It is not very practical to go one by one on this and I suspect that I might not be able to do it without automating the process. This is a considerable amount of work and it would make sense to check that I am on the correct track before spending time on this.

---

One worry that this experience showed is that if given a concept ID to check and if the concept ID is an RxNorm attached to a commercial name of a drug. Then the DDD calculated would be only for this exact drug and not for all the commercial names attached to the equivalent ATC code.