OHDSI Home | Forums | Wiki | Github

Drug_exposure.quantity recalculation

Dear all,

Hope that I don’t duplicate any topic.
If I understand correctly drug_exposure.quantity must be consistent with drug exposure dates.
I would like to know your opinion on the following questions:

  1. If it is OK to recalculate drug_exposure.quantity = (daily dose * actual duration) and to drop source quantity, when the original prescription was stopped before the planned date (please see example below)?

  2. What would be your recommendations on populating drug_exposure.quantity for medications without end date/ days_supply information? As we populate drug_exposure_end_date = drug_exposure_start_date for such records, which option is the best?

    • drug_exposure.quantity = NULL
    • drug_exposure.quantity = source quantity as it is
    • drug_exposure.quantity = daily dose * 1 day

Example 1 for question 1:

  • start date = 2015-05-12
  • end date = 2015-09-08
  • discontinue date = 2015-06-01
  • duration = 30
  • quantity = 60
  • refills = 3
  • frequency = 2 times a day
  • dose = 1

As discontinue date is before end date, then the following entry could be created in Drug_exposure:

  • drug_exposure_start_date = start date
  • drug_exposure_end_date = discontinue date

As quantity and days_supply must be consistent with the current exposure, it
can be calculated as following:

  • quantity = (drug_exposure_end_date - drug_exposure_start_date +1 day) * dose * frequency = 42
  • days_supply = drug_exposure_end_date - drug_exposure_start_date +1 day = 21
  • refills = NULL

Please share any experience of yours when you needed to readjust source quantity and duration values and which formulas/approaches you applied.

Hi @Alexdavv, @aostropolets, @Christian_Reich, @Vojtech_Huser, @MPhilofsky ! I saw you giving answers on similar topics. Could you please share your opinion on the questions above?

Many thanks in advance!

You need to put the most accurate values into these fields. So, if there is a quantity, but you happen to know that the actual quantity was less then use the latter information.

Don’t. Only populate what you have. If you have frequency and daily dose (sig) and duration information - that’s all that’s needed. The analytics will calculate the remaining. Only the _end_date should be inferred.

Bottom line: We routinely use several different heuristics to calculate duration and daily dose for drug exposure. @anthonysena is actually working on it. For the duration, we can use start and end date, days_supply, quantity * frequency or package size * frequency. The dose we can calculate from the strength or, if compounded or only ingredient / drug form given, the quantity information.

In what cases should we populate verbatim_end_date?
Should I always populate this field except for the cases when I guess the end date (for example, use 30 as default)?
And why do we need 3 fields: days_supply, drug_exposure_end_date and verbatim_end_date?
For example, there is a prescription: take the drug for 10 days started from 2020-01-01. How should I populate drug_exposure?
drug_exposure_start_date = 2020-01-01
days_supply = 10
drug_exposure_end_date = 2020-01-10
verbatim_end_date = 2020-01-10 or NULL?

Another example: we know that it is prescription with start date = 2020-01-01
drug_exposure_start_date = 2020-01-01
days_supply = 30 or NULL?
drug_exposure_end_date = 2020-01-30
verbatim_end_date = NULL

and for Tatiana’s example: should we recalculate days_supply and quantity? According to the wiki, days_supply ‘reflects the intention of the provider for the length of exposure.’
And a quantity of drug ‘as recorded in the original prescription or dispensing record’.

So, for example, a provider gave a prescription ‘take a drug (1 tablet once a day) for 10 days started 2020-01-01’. For some reason, the patient stopped taking the drug on 5th day:
start date = 2020-01-01
daily dosage = 1 tablet
quantity = 10 tablets
days_supply = 10
discontinue date = 2020-01-05

Should we recalculate days_supply and quantity? If we recalculate them, then they won’t reflect the intention of a provider for the length of exposure and quantity of drug as recorded in the original prescription.

Just in case when it’s directly defined in the source as a date. If you calculate/impute the end_date, leave this field empty.

drug_exposure_end_date - the real end_date of the exposure or the best assumption you could have. This will be primarily used for dose_era construction unless it’s NULL (what is out of the current constraints).

days_supply reflects the intention for the length of exposure. Even though the exposure was stopped, use the planned duration. May be used in dose_era construction when end_date is NULL somehow.

verbatim_end_date - just to reflect the source and probably for reference (e.g., to distinguish between real and inferred end_date).

I believe it’s NULL.

Why do you want to calculate this? I would prefer NULL.

Quantity - yes, since it should be used in dose_era calculations/analytics and indicate the real exposure.
We obviously need to make conventions more clear.

If we will drop the prescribed quantity and will calculate the actual quantity based on discontinue date: drug_exposure.quantity = (discontinue date - start_date + 1) * daily dose - it will contradict the definition of quantity field from OMOP CDM:

The quantity of drug as recorded in the original prescription or dispensing record.

  1. Would it be OK to populate drug_exposure.quantity with the value not from original prescription?

And one more question on discontinued/stopped medications :slight_smile:

  1. If I understood correctly from GitHub item End of a Drug Exposure · Issue #71 · OHDSI/CommonDataModel · GitHub : in the case of stopped/discontinued medications it is ok to have days_supply which doesn’t correlate with drug_exposure_end_date. As days_supply populated with the prescribed value may allow to analyse the prescribed duration with the real duration? Or should we recalculate days_supply based on discontinue date?

Brrrrrr. How about this:

days_supply - intention of the prescriber. Don’t recalculate because they stopped the exposure early.
quantity - actual quantity dispensed. Don’t recalculate because they stopped early.
verbatim_end_day: the discontinue day.

When you infer the drug_exposure_end_date, use in this order: verbatim_end_day - days_supply - quantity * daily dose - package size * daily dose - some default value (30 or 1 depending what drug it is, and if you don’t know just do 30). The end_date MUST be populated.

I’d like to add to the above.

If the drug is an “administered” drug, have end date = start date

t