Comment:

This specification requires the SMART on FHIR app to behave in a very specific way. I recommend making the currently specified approach a SHOULD or MAY, with its own conformance profile (i.e., so those that support the need this way can do so in a common way), but having a larger conformance profile which allows the SMART on FHIR app to be implemented however the payor wishes. Specifically, I think FHIRQuestionnaire + CQL + FHIR queries to pull data are good, but I don't think payors should be forced to use this approach. From an end user perspective, as long as it's a CDS Hooks service call followed by a SMART on FHIR app, there's no real need for the knowledge representation approach within the SMART on FHIR to be prescribed. What if, for example, the payor wished to highlight questions that it is confident is right according to the data pulled, as likely not needing manual review, and use a different highlight for questions that it is not so confident about, and wants to ensure is manually reviewed? What if the payor wishes to show the relevant data points that led to a conclusion? Or to allow the user to select from different data points, like SpO2 values, where perhaps the most recent is above the cutoff, but that was because it was while the patient was on O2 which was not captured, and the second to last one should be used instead? The current approach does not seem to allow this type of innovation and optimization to be allowed by payors.

Summary:

Encourage use of Questionnaire + CQL in the SMART on FHIR app, but allow other approaches as well.