I'm creating this ticket out of the comment from CQM-6653 so that it can get more visibility:

Thank you for the reply! The first part and most of second part of the answer makes total sense to us. Our understanding for patient-based measure is also that every clinician will get credit if some clinician takes the numerator action. In other words, for a patient-based measure, a patient will either be in the numerator or not, depending on whether they received proper numerator action. Who is attributed to the patient in the initial population should not have an impact on the patient's outcome. I'm glad we are on the same page.

For CMS-50 though, it is contradicting with what you said previously. You just said the CQL logic is not able to assign attribution to the provider, but you also said CMS-50 is clinician specific.

Say there is provider 1 who made referral 1 and 2, provider 2 who made referral 3 and 4. The referrals are in chronological order and the providers both have a qualifying encounter with the patient.

What would make sense to us is that

Either

• Provider 1 gets credit based on whether referral 1 has loop closed
• Provider 2 is not attributed to the patient because they didn't make the first referral for the patient

Or 

• Provider 1 gets credit based on whether referral 1 has loop closed
• Provider 2 gets credit based on whether referral 1 has loop closed, because referral 1 is the first referral on the patient's EHR and there is no CQL logic to assign attribution of this referral to a provider

Either way the patient should get an outcome solely based on the first referral on their EHR with the time frame. 

CQM-6625 and this ticket seem to suggest that 

• Provider 1 gets credit based on whether referral 1 has loop closed
• Provider 2 gets credit based on whether referral 3 has loop closed

If we can agree on what a patient-based measure means (which you seemed to do so in your first part of the response), this is clearly not a patient-based measure because there is no patient-level outcome.

Maybe we can also reach a consensus that what you are suggesting is actually very similar to other episode-based measures like CMS-68. In CMS-68, we are trying to make sure each encounter has medication documented; while for CMS-50, we are making sure each referral loop is closed. The common theme here is the outcomes are episode-based. With that, you can enforce each episode to be clinician specific if that is the measure intent.