Comments
Modifying attributes modify meaning in a finite number of predictable ways. For example, "entered in error" nullifies the resource and says the resource should not be considered for clinical purposes. Other elements and values reverse the meaning of the resource, such as "did not occur" or "refuted." Still others change the mode of a resource, for example, when the status is proposed or planned. Other values do not actually modify the semantics of the resource at all, but simply qualify the meaning (e.g. in-progress | completed or provisional | working | confirmed). Currently, no semantics are attached to modifying values. It would be much more powerful and computable if the values of a modifying attribute were marked as non-modifying, nullifying, negating, or mode-altering. Then a system processing the resource would know immediately what change of meaning (if any) was happening to the resource. Such meta-data concerning a modifying value could be quite powerful in search; for example, a search could exclude all negated resources without necessarily having to delve into the specific negating values on a resource-by-resource basis.

Grahame's Comments
We could do this in mappings on the codes. Suggest that Mark and I work on this

Disposition
Not Persuasive

Disposition Comment
The primary purpose of isModifier is simply to flag elements as "not safe to ignore if unrecognized". More sophisticated understanding of the various ways in which modification could occur (which is completely unconstrained and likely unenumerable) is not part of core. This could be done using extensions and documented using profiles in situations where this would be helpful.