HCC coding

The CMS-HCC model does not pay for diagnosis codes one by one. It sorts them into condition categories, a model most heavily shaped by the Medicare costs of chronic disease, and places related categories in hierarchies ranked by severity and cost. When a member has several conditions in the same hierarchy, payment is based only on the most severe and costly one; conditions in unrelated hierarchies add to each other. CMS publishes the ICD-10 to HCC mapping with its model software for each payment year.

Accuracy cuts both ways. A code needs support in the note, from a visit with an acceptable provider type, or it is a payment the plan cannot defend. A condition that is supported but never coded understates the member’s risk. Specificity often decides the category: an unspecified code can map to a lower category or to none, while the code for the same disease with its documented complication may land higher.

The mistake is treating HCC work as a year-end project or as a fixed list. Plans have to report each diagnosis at least once in every data-reporting period, so last year’s conditions do not carry forward on their own. The mapping also changes with the model in force: CMS finished phasing in its 2024 model in 2026 and is keeping it for 2027, so a crosswalk built for an older model gives wrong answers. Code from the current year’s mapping, and review open chronic conditions before visits rather than after the period closes.