Stream: implementers
Topic: Medication Dosing no drug.
Bill Lush (Mar 11 2021 at 17:39):
We are looking at how to prescribe contraceptive tablets, as the contraceptives are prescribed in high volumes we want to be sure we are getting it right.
So my question is we can prescribe a contraceptive as 1 tablet - once a day - for 3 weeks, we are OK with this. It is how do you specify having a week off from the contraceptive before starting the routine again?
Any guidance on how other systems are doing this would be much appreciated.
Thanks
Lloyd McKenzie (Mar 11 2021 at 17:56):
@Melva Peters @Jean Duteau
Jean Duteau (Mar 11 2021 at 18:24):
I've seen in some implementations a Timing.code of 21/28 which was 21 days out of 28 days. I think you can reproduce that with a frequency of 21 and a period of 28 days. And then text on your Dosage to indicate.
Lloyd McKenzie (Mar 11 2021 at 18:39):
This would be a good one to add to the list of pharmacy example dosages
Lloyd McKenzie (Mar 11 2021 at 18:39):
(Speaking of, tried to find a link to that set of examples from the Dosage data type, the Medication module and MedicationRequest - and couldn't see a link any of those places?)
Jean Duteau (Mar 11 2021 at 18:41):
care to make a JIRA issue? :)
Jean Duteau (Mar 11 2021 at 18:42):
Lloyd McKenzie said:
(Speaking of, tried to find a link to that set of examples from the Dosage data type, the Medication module and MedicationRequest - and couldn't see a link any of those places?)
http://hl7.org/fhir/medicationrequest-examples.html
We should change the text to actually have a link.
Elliot Silver (Mar 11 2021 at 18:45):
Jean Duteau said:
I've seen in some implementations a Timing.code of 21/28 which was 21 days out of 28 days. I think you can reproduce that with a frequency of 21 and a period of 28 days. And then text on your Dosage to indicate.
@Jean Duteau , stepping back a little from this specific issue, I keep seeing questions about various complex medication timings. The emphasis is always on how to express them, and we usually end up with a way to say it. Have we considered the other side of the equation? Do we expect consuming applications to be able use these convoluted timings in some automated fashion? While it is great to be able to say things in this "language," it needs to be understood on the other side.
Melva Peters (Mar 11 2021 at 18:48):
I didn't think it was possible to include examples in the Dosage data type so all of the examples are in MedRequest. Ideally the examples should be in the Dosage type...if that's possible...happy to put them
Lloyd McKenzie (Mar 11 2021 at 18:52):
You can't create examples of the Dosage type itself (except maybe using Parameters?), but you can certainly include hyperlinks from that page to wherever you've got example fragments maintained
Jean Duteau (Mar 11 2021 at 18:52):
@Elliot Silver totally agree. I feel that, in most cases, these complex or even mildly complex timings should fall back on text. Although Dosage can be made to say almost anything, there is a valid concern that receiving systems won't understand those representations.
Morten Ernebjerg (Mar 11 2021 at 21:02):
I also second the point @Elliot Silver made. We are currently looking at visualizing dosing information and a major challenge is just being able to cleanly identify and pick out the sort of instruction that can actually be displayed in a UI that is half-way readable. However, we work in in a multi-language setting, so free text is not great for us, either. Insofar as there are actually specific, more complex timing schemes that are commonly and broadly used, the easiest for us would arguable be to include them in the set of standard codes used for Timing.code (or have other large, but widely shared ValueSets for this element). The existing codes already overlap with timings that can be "atomically" expressed (using period, frequency etc.), there is even a nice table in the Timing documentation showing the correspondence.
Jean Duteau (Mar 11 2021 at 21:50):
Yes, and there is a statement in the Timing.code definition that says that if a code is present, it is expected to be a coding of whatever is in the structured data.
Morten Ernebjerg (Mar 12 2021 at 13:01):
Yes, exactly - that means that if I understand the code, I don't have to worry about details hiding the atomic definition.
Last updated: Apr 12 2022 at 19:14 UTC
