Per the HQMF Utils code found at https://github.com/pophealth/hqmf2js/blob/master/app/assets/javascripts/specifics.js.coffee, quote:

1. this is a little odd, but it appears when we have a negation with specifics we can
2. ignore the logical result of the negation. See comment in intersectAll.
3. we need to verify that we actually have some occurrences ...
4. this is a little odd, but it appears when we have a negation with specifics we can
5. ignore the logical result of the negation. The reason we do this is because we may
6. get too many negated values. Values that may be culled later via other specific
7. occurrences. Thus we do not want to return false out of a negation because the
8. values we are evaluating as false may be dropped.
9. we need to verify that we actually have some occurrences

This "always return true if there are specifics" feature of the code is preventing some patients from correctly falling into the numerator for CMS 114. I am not allowed to attach files to this ticket. We have supporting Bonnie patients available upon request.

If a patient has Occurrence A of a medication during Occurrence A of an encounter then the patient has specifics. During the AND NOT there is a negation and because there are specifics, the AND NOT is incorrectly evaluated as TRUE instead of FALSE.

e.g. and not ( A and B ) => and not ( True and True ) => and not (True) => is false but Cypress is evaluating it as true because of the above behavior of unionAll with specifics.