I am developing a demo application that acts like a middleware for transmitting data from point-of-care medical devices. For example, lets say you have a patient monitor that is generating lots of data like EKG, HR, SpO2, etc. My application will be able to capture that data and transmit it to a recipient using FHIR.
My application will know the location of the patient monitor and other details (observation time, value, etc) but because the interfaces to these patient monitors is often limited, my application probably won’t know what patient is attached to the patient monitor (ie the subject of the observation).
Is there some standard way to transmit observations using FHIR if you don’t know the patient ID?
It’s legitimate to transmit an Observation without a subject to an intermediary that will inject the subject (and possibly other elements such as normal range, basedOn, etc.). The key thing is that the subject gets identified before the data propagates to a general repository where knowing the subject will be essential to making use of the data. (Even if data gets anonymized, it’ll still have a subject to allow distinguishing that multiple observations were made on the same individual, even if minimal or no details about the individual are available.)
Is there any standard way to reference an anonymous subject? This is relevant because my application will probably mostly be used in research, where de-identifying is common.
You’ll still point to a Patient instance, but the instance will have less information. Perhaps just a gender and birth year, or an extension on the birthdate that indicates “age at enrollment” or something like that. You might also have an internal study-assigned identifier that could, if necessary, be de-referenced using out-of-resource information to determine who the individual actually was if consent and protocol allowed. All of the Observations, Procedures and other information would then point to that Patient instance.
