Introduction To DrugExposureDiagnostics

Drug(s) of interest

In the DrugExposureDiagnostics package, all the diagnostics are conducted on ingredient level, and if requested, additionally on drug level. We will use the ingredient “acetaminophen” as an example.
Here is a brief look at this ingredient.

Running the diagnostics

We can run all available checks at the same time using the ´executeChecks()´ function. This will return a list which contains the results of each check.
Multiple ingredients can be added in the vector of ´ingredients´

all_checks <- executeChecks(cdm,
  ingredients = c(1125315),
  subsetToConceptId = NULL,
  checks = c(
    "missing", "exposureDuration", "type", "route", "sourceConcept", "daysSupply",
    "verbatimEndDate", "dose", "sig", "quantity", "diagnosticsSummary"
  ),
  minCellCount = 5,
  sample = 10000,
  tablePrefix = NULL,
  earliestStartDate = "2010-01-01",
  verbose = FALSE,
  byConcept = TRUE
)
#> population after earliestStartDate smaller than sample, sampling ignored
#> ℹ The following estimates will be computed:
#> • daily_dose: count_missing, percentage_missing, mean, sd, q05, q25, median,
#>   q75, q95, min, max
#> ! Table is collected to memory as not all requested estimates are supported on
#>   the database side
#> → Start summary of data, at 2024-11-23 16:43:29.257437
#> 
#> ✔ Summary finished, at 2024-11-23 16:43:29.72384
#> `sample_size` casted to character.
#> `sample_size` eliminated from settings as all elements are NA.

The cdm is the database reference of the OMOP CDM using the CDMConnector package.
The ingredients is a list of ingredients of interests, by default it is 1125315 for acetaminophen. The subsetToConceptId vector of concept IDs of the ingredients to filter. If a concept ID is positive it will be included, a negative one will be excluded. If NULL, all concept IDs for an ingredient will be considered. checks allows to select the checks to be executed, by default the missing values, the exposure duration and the quantity checks will be run. The minCellCount is minimum number of events to report, numbers lower than this will be obscured. sample is the number of samples, by default, 10.000 drug record samples will be used.
The tablePrefix is an optional value for database tables that will be created during executeChecks. earliestStartDate is the earliest data from which drug records will be included. verbose is a parameter that enables the printing of messages to the console. byConcept is a boolean that determines if only overall results should be returned or also by drug concept.

The check “missing” outputs the missingValuesOverall if byConcept = FALSE and additionally missingValuesbyConcept if byConcept = TRUE
The check “exposureDuration” outputs the drugExposureDurationOverall if byConcept = FALSE and additionally drugExposureDurationbyConcept if byConcept = TRUE
The check “type” outputs the drugTypesOverall if byConcept = FALSE and additionally drugTypesbyConcept if byConcept = TRUE
The check “route” outputs the drugRoutesOverall if byConcept = FALSE and additionally drugRoutesbyConcept if byConcept = TRUE
The check “sourceConcept” outputs the sourceConceptsOverall (which is actually the drug concept level. It disregards the byConcept argument because only the drug concept level makes sense here)
The check “daysSupply” outputs the drugDaysSupplyOverall if byConcept = FALSE and additionally drugDaysSupplybyConcept if byConcept = TRUE
The check “verbatimEndDate” outputs the drugVerbatimEndDate if byConcept = FALSE and additionally drugVerbatimEndDatebyConcept if byConcept = TRUE
The check “dose” outputs the drugDoseOverall (output only on ingredient level. It disregards the byConcept argument because only the drug concept level makes sense here)
The check “sig” outputs the drugSig if byConcept = FALSE and additionally drugSigbyConcept if byConcept = TRUE
The check “quantity” outputs the drugQuantity if byConcept = FALSE and additionally drugQuantitybyConcept if byConcept = TRUE
The check “diagnosticsSummary” outputs a summary called diagnosticsSummary

Here, we see the files created:

names(all_checks)
#>  [1] "conceptSummary"                "missingValuesOverall"         
#>  [3] "missingValuesByConcept"        "drugExposureDurationOverall"  
#>  [5] "drugExposureDurationByConcept" "drugTypesOverall"             
#>  [7] "drugTypesByConcept"            "drugRoutesOverall"            
#>  [9] "drugRoutesByConcept"           "drugSourceConceptsOverall"    
#> [11] "drugDaysSupply"                "drugDaysSupplyByConcept"      
#> [13] "drugVerbatimEndDate"           "drugVerbatimEndDateByConcept" 
#> [15] "drugDose"                      "drugSig"                      
#> [17] "drugSigByConcept"              "drugQuantity"                 
#> [19] "drugQuantityByConcept"         "diagnosticsSummary"           
#> [21] "metadata"

Assessing the drug concepts (of interest)

The first item shown, the conceptSummary is not the output from a check but always output. This table is used within the function as an input for all the requested checks.
It contains information from the CONCEPT table and DRUG_STRENGTH table.
The conceptSummary has several use:
It can be quickly seen whether a certain drug is mapped on ingredient level. Thereby, the drug_concept_id and ingredient_concept_id are identical.
Furthermore, the conceptSummary can be used to make decisions about further refine the concept list. For example, though the argument subsetToConceptId in the executeChecks() function, the selected concepts can be requested or eliminated.
The conceptSummary in the case of acetaminophen with subsetToConceptId = NULL, the following drug concepts are part of the diagnostics investigation.

datatable(all_checks$conceptSummary,
  rownames = FALSE
)

Saving the diagnostic output

After running the checks, we can write the created outputs (CSV files) into a zip file to disk using the writeResultToDisk() function. The zip created from using the writeResultToDisk() function serves as the input for the corresponding Shiny Application.
Please replace the “your_database” with the name of your database. Create your designated “output_folder” at your desired location and put the path if not the working directory.

writeResultToDisk(all_checks,
  databaseId = "your_database",
  outputFolder = "output_folder"
)

View results in the Shiny app & make it available for publishing

viewResults(
  dataFolder = file.path(getwd(), "output_folder"),
  makePublishable = TRUE,
  publishDir = file.path(getwd(), "MyStudyResultsExplorer"),
  overwritePublishDir = TRUE
)