Build a measure inventory your team can trust
Understand CMS measure data sources, eligibility, exclusions, cut points, and the difference between Part C, Part D, survey and operational measures.
Give every measure a durable record
Include program, year, measure name, source, population definition, calculation owner, and operational owner. Treat a similar display name as a prompt to investigate, not proof that two definitions match.
Trace the inputs before the result
For each input, record where it originates, when it arrives, and what validation it passed. Keep the accepted file date next to the measure run date. This turns a reconciliation discussion into a comparison of known inputs.
Separate missing from not applicable
Use distinct states for unavailable data, excluded observations, incomplete calculation, and a genuinely zero result. Give each state a next action. A blank that disappears in a dashboard can become an incorrect decision.
Reconcile with an exception register
Compare internal and official or vendor results at the smallest permitted level. Log differences, their owners, the investigation, and disposition. Do not clear a provisional label simply because the total looks plausible.
Take this to your next review
Bring a dated source, the assumption you are using, the person responsible for the next action, and the evidence you will review. Keep official methodology separate from local planning choices.
Reference desk
These resources support the program context. The workflow above is AdherenceStar’s suggested practice, not a CMS-mandated procedure.
CMS · 2026 technical notes ↗CMS · Part C and D performance data ↗