ICD-10-CM Diagnosis Coding for Risk Adjustment
Specificity drives capture
ICD-10-CM codes are alphanumeric and built for specificity: category, etiology, anatomic site, laterality, and a seventh character for encounter type or healing status. Risk adjustment capture depends on the most specific code supported by documentation, not the shortest code available.
Combination codes bundle a systemic disease with its manifestation - diabetes with CKD (E11.22 plus N18.32), diabetic retinopathy, hypertensive heart disease - and should replace separate, less specific codes. The ICD-10-CM "with" convention assumes a causal link in listed combinations, but documentation must still support both conditions. External cause codes, most Z codes (history, screening, status), and symptoms without a confirmed diagnosis generally do not map to HCCs. "History of" codes describe resolved conditions and are not active diagnoses; sequela codes may be captured when the residual effect is being treated. Never code "rule out," "probable," or "suspected" in outpatient risk adjustment.
Traps: using an unspecified code when a stage or type is documented; splitting a combination code into two less specific codes; and coding a history code as if the disease were active.
Rule of thumb: code to the highest documented specificity and use combination codes first. Worked example: documentation says "type 2 diabetes with stage 3a CKD"; report E11.22 and N18.32 rather than E11.9 and N18.9.