Three parts to this question.
My assumption is that for every MedicationDispense resource that a pharmacy creates, there would be a corresponding MedicationStatement. There may be multiple MedicationDispense resources associated with a single MedicationStatement (e.g. refills). Is this consistent with the view of others in the community?
The spec indicates that MedicationStatement is a less formal MedicationAdministration. In the case where MedicationAdministration is created, should there always be a corresponding MedicationStatement? Again, my assumption is yes, where MedicationStatement simply is an indicator that a Patient is supposed to be using a particular medication, while MedicationAdministration is a very specific event that the medication was actually administered to the patient.
For self (patient) reported administrations, is MedicationAdministration recommended? If not, this seems to cloud the use of MedicationStatement, where it is used for both the general (“I’m supposed to be taking this medication”) and the specific (“I took my dose this morning at 7 am”). Thoughts?
There wouldn’t necessarily be any MedicationStatements associated with a MedicationDispense - some environments might not use MedicationStatement at all. In other cases, a single MedicationStatement might span a full treatment regimen covering multiple prescriptions and numerous dispenses. MedicationStatement is a “summary” resource. The granularity at which the summary is captured can vary and will depend on business need.
Rather than “supposed to be using”, I’d rephrase it as “believed to be using”. (They may in fact be not supposed to be using - as MedicationStatement can be used to record the use of non-prescribed and illicit substances as well as therapeutic products - whether prescribed or not.) The belief may come from the existence of prescription, dispense or administration records, from lab results or from the statement of the patient or their care-givers.
In general, MedicationAdministration makes sense to capture individual administrations - it has the appropriate attributes and is intended for this level of detail. The record allows you to clearly indicate who performed the administration. If you care about who recorded the information, Provenance would support this.
Thanks for your thoughts, as well as the clarifying statement around “supposed to be taking” versus “believed to be taking”, I agree with that.
Regarding the use of Medication Statement, while I agree that there’s no guarantee that a particular pharmacy (or other partner for that matter), may use Medication Statement, is there a preference in the community for how a medication history should be provided? That’s the context I run into most often. As a consumer, if you had a choice in how medication history would be obtained from a pharmacy partner:
- Collection of Medication Statement
- Collection of Medication Order
- Collection of Medication Dispense
- (2) and (3) above
- All of the above
My opinion: I think the answer is likely to vary depending on who’s looking at the medication history and what they care about. If compliance is a concern/consideration, then knowing the orders and dispenses is useful. If you just care about “what meds is the patient on”, then the lower-level artifacts are just noise. In the US, I believe Argonaut has settled on primarily representing meds using MedicationStatement, but in practice I think you’ll see a variety of outputs, depending on the capabilities of the systems and who they expect to talk to.
Other opinions welcome :>
Correct. The Argonaut implementation guide uses MedicationStatement for medication history.
We don’t have a lot of experience with folks implementing MedicationAdministration/MedicationDispense.
I’ll just add that the Pharmacy work group had a lot of recent discussions about how to interpret the combination of MedicationStatement.status in conjunction with MedicationStatement.taken. Those discussions led to a new table, http://build.fhir.org/medicationstatement.html#126.96.36.199, showing what combinations are possible. For example, it is possible to have an active status with taken = no, which would mean that the patient “should be” taking the medication (active), but did not comply (not taken). The table is trying to reflect that compliance (i.e. MedicationStatement.taken) is only documented/relevant for “active” (active, on hold, intended) outpatient prescriptions. The MedicationStatement.taken is not applicable for an inpatient med that a practitioner administered or an order that has been end-stated.
This actually led into a much bigger discussion (deferred until after STU3) whether we needed to decouple the med history list (which includes both orders/requests + patient reported meds) from compliance (taken or not). Currently, in the US, MedicationStatement is trying to meet all of the following:
- MedicationStatements derived from MedicationRequests – needed for complete med history
- MedicationStatements based on a statement from the patient/related person – needed for complete med history
- Compliance of whether the patient is taking the medication (ordered or otherwise) as expected