Modelling zibs that are part of a CarePlan

[pieter-edelman-nictiz]

Zibs NursingIntervention, HelpFromOthers and CareAgreement correlate with the FHIR-resource CarePlan, but the mapping is not direct; rather, these zibs seem to be about CarePlan.activity rather than an entire CarePlan. In other words, when talking about a CarePlan, one expects is to be made up from these zibs rather than representing these zibs directly. This issue is somewhat compounded by the observation that NursingIntervention and HelpFromOthers severely overlap (see https://bits.nictiz.nl/browse/ZIB-2027)

Up until we now, we kind of ignored this issue and made profiles where one could only instantiate a CarePlan with a single .activity. However, this might run counter to how FHIR is used in practice and is also not really in line with the intention of FHIR. Modelling these zibs as an .activity leads to several questions however:

• It requires the information standards where these zibs are used to describe how instances of these zibs relate to each other, and to care plans.
• There is no obvious way to profile just an .activity -- it is neither a resource nor a datatype. Options would include:
  • Create a "generic CarePlan profile" where each of the three zibs is mapped onto some slice (how these should be discriminated is an open question). This could than form the raw material that a derived standard could slice and dice to suit its needs.
  • Create CarePlan profiles for each of these zibs but leave everything outside .activity alone. CarePlan-instances could claim conformance to more than one of these profiles if it represents more than one zib. This is probably more tricky to model in an information standard.
  • Even so, there are still issues with comments and dates that don't cleanly map to .activity. This could be solved with extensions if there are no other options.

[pieter-edelman-nictiz]

In R5, the .activity is mostly emptied out except for a reference field to other Request resources. This field is already present in R4 (.activity.reference). Maybe it would be better to make profiles on ServiceRequest or something like that?

[pieter-edelman-nictiz]

Some notes:

• zib NursingIntervention, when interpreted as a plan, seems to fit really wel with both the intent and the model of a ServiceRequest. Wouldn't this be a better model than CarePlan anyway?
• zib CareAgreement fits somewhat to a ServiceRequest. The parts that match poorly are AgreementParty, Activity (as string) and Explanation (as string). But this is true for the CarePlan as well.
• zib HelpFromOthers might match in intent to a CarePlan, but not at all in modelling. It mainly seems to be about the question who is involved around an activity rather than the fact that the activity needs to be done. So it has some aspects of a CareTeam, but then focused about some particular activity.

[pieter-edelman-nictiz]

Zib HelpFromOthers might even be an Observation, combined with a CareTeam. From ZIB-2033 it can be gathered that the intent is to provide a statement about the help a patient currently receives for certain situations. So it's neither about actionable activities nor about the team, just the notion that help is available for some activities.