Stream: implementers
Topic: Representing alerts etc. from patient
Jon N. Didriksen (May 07 2021 at 11:45):
I work with a solution where patients live at home and have sensors that reportmeasurements (blood pressure, heart rate, etc.), as well as a tablet where they fill out questionnaires. The system generates warnings to health care workers if the measured values are outside the normal range, if the score calculated from a questionnaire response is above/below certain limits or if tasks have not been performed (measurements, questionnaire responses etc.). When these alerts are followed up and closed, they must be reported to the EHR, and we use FHIR for this exchange. For now, the information ends up as a text note in the EHR, but we want to represented the information in a structured way, because the same structures will later be used in other contexts (among other things, it is relevant to share measurement data with a doctor).
The biggest question is how the alerts should be represented. It looks like Flag is the resource that fits best, but this is not completely clear. Opinions?
So far, this is what we plan for the complete representation:
Measurement data = observation, vital signs profiles
Form answer = QuestionnaireResponse (for now, the form definition - Questionnaire - is not relevant). The form has a score and limit values for yellow/red alerts, which we plan to represent as a related Observation.
Note describing follow-up actions: Composition.
Typically we will then send in a Bundle containing the alert (Flag?), detailed information about what the alert was about (Observation, QuestionnaireResponse, or no additional information - if the alert is due to lack of action), note describing follow-up (Composition).
Comments and suggestions are appreciated!
Lloyd McKenzie (May 07 2021 at 13:21):
I think of 'Flag' as a yellow-sticky on the patient folder that anyone dealing with the patient should read before they interact with the Patient. It generally summarizes 'high importance' information that's also represented with a different resource. You could use this to draw attention to your desired action, but it wouldn't be the 'primary' representation of the desired action.
You could use DetectedIssue with mitigation, but there's no clear ownership/responsibility for management of those. The other possibility would be to assign a Task to a practitioner, care team or organization. The Task would be "manage/follow-up on this result".
Dag Høidahl (Dec 02 2021 at 14:43):
(deleted)
Dag Høidahl (Dec 02 2021 at 14:44):
(deleted)
Dag Høidahl (Dec 02 2021 at 14:54):
@Lloyd McKenzie Sorry to bring up this topic again. I'm confused about your suggestion to apply DetectedIssue in this context, so I hope you can help me understand it better. From the documentation, DetectedIssue is "a concern about an existing or proposed set of clinical activity". In another thread[1], you said:
DetectedIssue represents a warning about a potential contraindication.
The case here is not about whether the clinical activity (the measurements or questionnaires) should be changed, but how to represent what a clinician did when the results of the clinical activity indicated a problem. I understand that Flag doesn't cover this case particularly well, but I fail to see why DetectedIssue does, given my understanding of the documentation and your quoted response. Could you help me sort where I've misunderstood?
Lloyd McKenzie (Dec 02 2021 at 17:21):
Detected issue includes both the issue and the mitigation of the issue.
Last updated: Apr 12 2022 at 19:14 UTC
