In CMS specs for the QRDA Cat 3, for Measure Data (observation: templateId 2.16.840.1.113883.10.20.27.3.5) it requires an observation node which contains aggregate count for a specific population (e.g. numerator, denominator, etc). It seems that this count is based on person level, which means it's just the number of patients qualifying for that population. However, if this is a visit-based measure, do we still count the number of patients instead of encounters? If we still count patients, how do we determine to which population one patient belongs? For instance, if a patient has multiple encounters (some are qualifying for denominator exclusion, some are qualifying for numerators), is this patient qualified for denominator exclusion, numerator or both?