Prepare for compatibility with CH IPS

[dvribeira]

With the preparation of the swiss IG of the International Patient Summary, there are some issues that should be identified in order for getting the PMP ready for this.

In principle, I see no use in providing other resources than the PMLC medication statements for the summary, hence it should be enough to be sure that at least these resources are compatible. Obviously other infrastructure resources like Patient an so on will also have to be checked.

For now, the CH IPS profile medication statement's structure imposes only cardinality constraints on making the medication and the subject mandatory. That's all right.

CH IPS medication statement profile also imposes conformance with FHIR's MedicationStatementIPS. After a fast look this might pose issues with several elements:

• effective[x] (dateTime/period) is must support with 1..1 cardinality. We do not support it.
  • data-absent-reason is must support. <- added extension
• dosage.text is must support <- we should probably also make it must support
• dosage.time is must support <- idem

[dvribeira]

The biggest problem seems to be the effective[x] element. At first I can identify the following approaches:

• We change the PMP to use effective[x] for the treatment's start and end date. This a deep change, will considerably impact us and integrating systems. Furthermore there's probably good reason why the dosage boundsPeriod was chosen instead.
• We ask to provide both effective[x] and dosage boundsPeriod for medication statements. There's no use in providing both, integrators must also adopt this change.
• We create a new derived profile for medication statement for IPS exports or we simply modificy the PMLC med statement, including the effective[x] as extracted from the consolidated boundsPeriod. In this case it's only the aggregator that would have implement changes and they are simple. The information would still be redundant. Modifying the PMLC med statement would be the best approach, since IPS summaries would not be created by the aggregator, and it would allow whatever module creating them jut fetch a PMLC from the PMP to get the statements needed for the IPS.

The first option would better align how we use our resources with FHIR. This in itself would be a good idea. However it implies the biggest effort on all involved parties. The second option would effectively solve the problem with minimum effort.