parseHealthCanadaDPD: Parse the Health Canada Drug Product Database (DPD) Data...

View source: R/health_canada_dpd_parser.R

parseHealthCanadaDPDR Documentation

Parse the Health Canada Drug Product Database (DPD) Data Extract

Description

Reads Health Canada's Drug Product Database data extract and returns a dvobject with its 13 tables. Each of the four product sets (marketed, approved, cancelled, dormant) is read from its unzipped folder or straight from its zip; the rows of every set go in one table with an extract column.

Usage

parseHealthCanadaDPD(
  dpd_dir,
  extracts = c("marketed", "approved", "cancelled", "dormant"),
  release = NULL,
  release_date = NULL,
  retrieved_at = NULL
)

Arguments

dpd_dir

Character. Folder holding the extract's zips or unzipped folders.

extracts

Character. Product sets to read, any of "marketed", "approved", "cancelled" and "dormant" (all by default).

release

Character or NULL. Release label, e.g. "2026-10". DPD publishes no release number.

release_date

Date, "YYYY-MM-DD" or NULL. Release date.

retrieved_at

Date, "YYYY-MM-DD" or NULL. Date the files were downloaded.

Details

dpd_dir holds the extract as Health Canada distributes it, each product set either as its zip or unzipped into a folder of the same name: allfiles (marketed), allfiles_ap (approved), allfiles_ia (cancelled) and allfiles_dr (dormant), plus inactive (the inactive products list). Zips are extracted to a temporary folder that is removed after reading. When neither inactive nor inactive.zip is there, inactive.txt is read from the cancelled set, which holds the same file.

The files have no header: column names are those of Health Canada's read-me ("Read Me File - Drug Product Database (DPD) Data Extract"), lower case, and table names are the read-me's without QRYM_. Every value is quoted and the files are UTF-8; empty values become NA. Code columns (drug_code, active_ingredient_code, company_code, pharm_form_code, route_of_administration_code, si_code) are integers and the DD-MON-YYYY dates are dates; every other column stays text as written. Dates are kept as written, including two lot expiry dates in the cancelled set written with years 0030 and 0031.

The therapeutic class files hold four fields where the read-me lists seven columns: the three certain ones are kept and the parser stops if the fourth ever holds a value.

The parser groups nothing: status holds every status row as written, with current_status_flag marking each product's current one.

DPD data are published under the Open Government Licence - Canada (version 2.0); the attribution it asks for is recorded in the object's metadata (see get_source_metadata). The DPD extract page also links to Health Canada's website terms; the dataset record on open.canada.ca states the Open Government Licence – Canada.

Value

A dvobject with 13 tibbles, each keyed by drug_code: drug_product, active_ingredients, companies, form, packaging, pharmaceutical_std, route, schedule, status, therapeutic_class, veterinary_species, biosimilars (each with extract) and inactive_products. inactive_products lists cancelled products, so it is built only when extracts includes "cancelled" and an inactive products file is found.

See Also

Other parsers: cett_nodes_options(), drug_node_options(), parseDrugBank(), parseHGNC(), parseHPO(), parseOnSIDES(), parseOpenFDA(), parseTWOSIDES(), references_node_options()

Examples

dpd_dir <- system.file("extdata", "health_canada_dpd", package = "dbparser")
dpd <- parseHealthCanadaDPD(dpd_dir, release = "2026-10",
                            retrieved_at = "2026-10-21")
dpd$drug_product[, c("drug_code", "brand_name", "extract")]
get_source_metadata(dpd)


dbparser documentation built on Nov. 9, 2026, 1:10 a.m.