Guidance needed for Drug Exposure days_supply field

[MelaniePhilofsky]

Issue Name: Guidance needed for Drug Exposure days_supply field

Issue Type: Lack of guidance

Description: The Drug Exposure days_supply field only gives guidance if the source data lack these data. Guidance isn't given if the data have negative values or unrealistically high values.

Related forum post: Linked

Reported By: Melanie Philofsky

[MelaniePhilofsky]

Maxim to sponsor

[MaximMoinat]

Find the current documentation on the drug.exposure.days_supply field below:

Days supply of the drug. This should be the verbatim days_supply as given on the prescription. If the drug is physician administered use duration end date if given or set to 1 as default if duration is not available.

Summary of issues:

• No guidance for negative values
• No guidance for unrealistic high values
• The drug_exposure_end_date is a mandatory field. This makes days_supply redundant as it can always be calculated from end_date - start_date + 1. Which field to trust when they differ?
• The advice to set as 1 by default should b revised. The days_supply can also be left empty as it is a non-required field. (even though a duration still has to be assumed for the drug_exposure_end_date)

Prior discussions on OHDSI forums:

• Days supply (yet again) (2018)
• Drug_exposure.quantity recalculation (2020)
  • Quoting Christian:

  days_supply - intention of the prescriber. Don’t recalculate because they stopped the exposure early.

• EHR Prescription data - best practice (2023)

[MaximMoinat]

Proposal after Themis WG meeting 2023-10-06

• Open PR with new descriptions: Update days_supply description and convention CommonDataModel#593

Update days_supply field description (currently empty)

The number of days of supply of the medication as recorded in the original prescription or dispensing record. Days supply can differ from actual drug duration (i.e. prescribed days supply vs actual exposure).

Update days_supply ETL convention

The field should be left empty if the source data does not contain a verbatim days_supply, and should not be calculated from other fields.��Negative values are not allowed. Several actions are possible: 1) record is not trustworthy and we remove the record entirely. 2) we trust the record and leave days_supply empty or 3) record needs to be combined with other record (e.g. reversal of prescription).
�High values (>365 days) should be investigated. If considered an error in the source data (e.g. typo), the value needs to be excluded to prevent creation of unrealistic long eras.

Study design recommendation

Calculate drug duration from drug_exposure_end_date and drug_exposure_start_date.

[clairblacketer]

This is ratified and added to the CDM documentation. We may decide that it will be moved to the THEMIS repo during April Olympians but I am closing this for now.