Stream: patient empowerment
Topic: patient entered medication
John Moehrke (Nov 24 2020 at 15:07):
Is there a Profile of how a patient entered medication record would be supported in FHIR R4? It seems a typical layout of information is not obvious how it falls into MedicationStatement, MedicationDispense, Organization (pharmacy), etc.
René Spronk (Nov 25 2020 at 08:19):
MedicationUsage (MedicationStatement) and MedicationAdministration would be the prime candidates, unless the patient happens to have a computable expression of the details one of the other Medication* resources, e.g. if they pulled that information from elsewhere.
Dave deBronkart (Nov 25 2020 at 13:31):
Hey @John Moehrke give me a quick scenario where this will be used? (I can imagine lots but I wonder how it lives in the FHIR world)
John Moehrke (Nov 25 2020 at 14:04):
Hi @Dave deBronkart as with any data the patient enters (Patient Generated Health Data - PGHD) the intention is to engage the patient. Many patients choose to take over-the-counter, herbal, or 'other' medications. These medications might impact treatment. The other use-case is where the patient has received care at a different facility, thus their records are not known to their primary care doctor. This might be because of emergency care, urgent care, dental care, etc... thus the whole point of PGHD is to enable the patient to inform their caregivers.
John Moehrke (Nov 25 2020 at 14:09):
Yes MedicationStatement is obvious to hold most of the information typically entered by the patient... but it is not for everything.
- prescription number, clinician who prescribed, pharmacy, and phone of the pharmacy
Often the patient doesn't know more than these. As it stands it looks like I have to create a contained MedicationDispense just so that I can have a MedicationDispense.performer.actor to hold a contained Organization where I put the name of the pharmacy, and their phone number
I am not sure where to put the prescription number, stuffing it in MedicationStatement.identifier for now
Lastly, the dose strength and frequency is very confusing. Seems I can say the same thing in many many different ways. What is the right way to say: taking two pills three times a day?
Michele Mottini (Nov 25 2020 at 14:21):
If this is in the US note that MedicationStatement has been dropped from US Core - so it won't be usually supported
John Moehrke (Nov 25 2020 at 14:43):
mandates related to US-Core are not relevant. If US-Core could guide, that would be relevant. I am looking for specification, not mandate.
John Moehrke (Nov 25 2020 at 14:49):
I expect that the patient empowerment has other PHR implementors that have had to address what I am looking to do. @Abigail Watson @Debi Willis others?
John Moehrke (Nov 25 2020 at 14:49):
wire diagram of a prototype UI image.png
Melva Peters (Nov 25 2020 at 17:57):
The Pharmacy Work Group would recommend that patient reported medications be added using Medication Usage.
John Moehrke (Nov 25 2020 at 18:19):
I don't see a MedicationUsage in R4
Peter Jordan (Nov 25 2020 at 20:31):
Michele Mottini said:
If this is in the US note that MedicationStatement has been dropped from US Core - so it won't be usually supported
What was the thinking behind that decision? AFAIA, MedicationStatement is in the ONC Common Data Model and it's also used in Argonaut (although that's not R4, of course) and the IPS (not based on US Core, but supported by ONC).
Michele Mottini (Nov 25 2020 at 21:16):
What was the thinking behind that decision
Not sure , but I think there was some overlap with MedicationOrder / MedicationRequest that caused problem (as a matter of fact we are getting lots of duplicate stuff now if we query for both MedicationOrder and MedicationStatement from existing Argonaut servers) (@Brett Marquard)
Dave deBronkart (Nov 25 2020 at 22:04):
John Moehrke said:
Hi Dave deBronkart as with any data the patient enters (Patient Generated Health Data - PGHD) the intention is to engage the patient. Many patients choose to take over-the-counter, herbal, or 'other' medications. These medications might impact treatment. The other use-case is where the patient has received care at a different facility, thus their records are not known to their primary care doctor. This might be because of emergency care, urgent care, dental care, etc... thus the whole point of PGHD is to enable the patient to inform their caregivers.
Sorry, @John Moehrke , I should have been more explicit about "where this will be used." I didn't say so but the question I was thinking about was - what's the situation where a patient enters something, vs it existing in some system?
I think I got hung up by "patient entered". But from context so far, I'm guessing you mean a patient has an app (or an Apple-style aggregation) that knows stuff (manually entered or e-collected from elsewhere) and wants to submit it to some EHR, yes? No?
John Moehrke (Nov 25 2020 at 22:06):
all the above... Patients know more than any one EHR.
Dave deBronkart (Nov 25 2020 at 22:07):
Is there any special meaning to patient "entered"?
John Moehrke (Nov 25 2020 at 22:08):
I have no idea what meanings that might have that is not the obvious... "The Patient enters the medication details". I can you offer some alternatives that you are thinking are in vs out?
John Moehrke (Nov 25 2020 at 22:10):
I am indeed working off of the HHS/ONC description under Patient Generated Health Data https://www.healthit.gov/topic/scientific-initiatives/patient-generated-health-data
But I am not thinking something that is not global in need.
Dave deBronkart (Nov 25 2020 at 22:10):
Well, I proposed (and you accepted in your "all the above" :-)) that an app might have medication data from elsewhere, e.g. perhaps downloaded from my pharmacy. I wouldn't call that patient "entered."
Dave deBronkart (Nov 25 2020 at 22:11):
John Moehrke said:
I am indeed working off of the HHS/ONC description under Patient Generated Health Data https://www.healthit.gov/topic/scientific-initiatives/patient-generated-health-data
But I am not thinking something that is not global in need.
Aha! So I get to refresh my ping of @Steve Posnack :smile:
John Moehrke (Nov 25 2020 at 22:12):
You are concerning yourself with a workflow that is important, but well beyond just getting the format consistent.
Dave deBronkart (Nov 25 2020 at 22:13):
This where Zulip needs the ability to fork a tangential conversation so it doesn't clog the thread. I'll clam up now.
John Moehrke (Nov 25 2020 at 22:13):
I just want to have a medication entered by a patient to be reliably encoded. If the patient uses an app to pull medications from somewhere in FHIR format, the I would like that pulled FHIR format to be compliant with some reliably encoded medication. If that app submits that data to some other organization, like their GP or outpatient or surgery center, I would like that medication to be in a reliably encoded medication form
John Moehrke (Nov 25 2020 at 22:14):
thus, I am more interested in reliably encoding a medication... once that is had, then I would worry about higher level flows. but it does not do much good to worry about the higher level flows with data that is not useful (aka the definition of Interoperabilty is important)
Melva Peters (Nov 25 2020 at 22:17):
@John Moehrke in R4 it is called MedicationStatement - was renamed to MedicationUsage in R5
John Moehrke (Nov 25 2020 at 22:17):
that is an unfortunate rename... is it really necessary to rename and break everything?
Melva Peters (Nov 25 2020 at 22:19):
We had lots of discussion about the name change before it was made including with the implementers who participate in Pharmacy
Lloyd McKenzie (Nov 25 2020 at 23:22):
If you're going to capture details about a prescription, you'll use MedicationRequest - even if you're not capturing much. The resource allows that to be reported. If you just want to say "I'm taking this med", you'll use MedicationStatement/MedicationUsage. I expect that US Core will eventually get to the point of supporting both MedicationUsage and MedicationRequest. There will be overlap in the data exposed, but it's at different granularity.
Peter Jordan (Nov 26 2020 at 23:58):
To generate a Medication Statement, the most useful information - if available - is the medication that the patient is actually taking (MedicationAdministration) and/or the medication given to a patient (MedicationDispense). As a practical example, whenever possible, downstream users of data from the NZ National e-Prescription Service look at the Dispense Records - partly because of generic prescribing v brand dispensing, but also because of the % of unfilled prescriptions. Other sources - in FHIR terms, MedicationRequest, Observation & Claim - might be described (as in the Spec) as 'supporting information'.
Brett Marquard (Nov 30 2020 at 11:32):
@Michele Mottini I am sorry to hear you are getting lots of duplicate stuff -- hopefully MedicationStatement.basedOn helps a little.
Historical context on the change from MS-->MR.
MedicationRequest allows an Order to still be active even though I am not taking it, which was a requirement the brought forward by several implementers. This isn't allowed in R4 MedicationStatement... This was allowed in R2/R3 (MedicationStatement.taken=N).
@John Moehrke Our work in Argonaut, and subsequently US Core, was developed in collaboration with the implementer community -- no regulators were at the table. I understand this still doesn't solve your quest of 'patient-entered' medications. We do give some small amount of guidance in US Core on using reported[x] on the MedicationList page.
The Medication resource overlap is tricky -- we focused on how to make sure a client would get all the active meds. If FHIR core wanted to make a Patient reported profile, or add lots of examples, that would be great.
John Moehrke (Nov 30 2020 at 14:09):
often the patient only has the bottle in their hand... so there is not much information to fill out a MedicationRequest. It feels down right wrong for patient entered prescription to record this as a MedicationRequest. Somehow recording a MedicationDispense feels less wrong, but it also seems wrong. These two resources have a clearly described agent that is expected to create them, during a clearly defined point in a workflow. The case where a patient is carrying a prescription bottle to their next health encounter is clearly not about either of these workflows.
In a perfect world, all prescriptions are well communicted to everyone that should see them... but in the real-world patients get prescribed medications at various points that are not as well connected.
John Moehrke (Nov 30 2020 at 14:11):
right now I have modeled a MedicationStatement with a contained MedicationRequest, MedicationDispense, and Organization. (FHIR R4)
Dave deBronkart (Nov 30 2020 at 16:41):
Melva Peters said:
We had lots of discussion about the name change before it was made including with the implementers who participate in Pharmacy
@Melva Peters as I continue learning about co-chair duties, it occurs to me to ask: are those discussions recorded somewhere so that an interested party can go back and review the thinking?
It always seems to be a big question - how much detail to capture about each side of a debate.
Melva Peters (Nov 30 2020 at 16:42):
the discussion is in the minutes of the Pharmacy WG. @Dave deBronkart
Last updated: Apr 12 2022 at 19:14 UTC
