How to Calculate RAF Scores: Step-by-Step Walkthrough with Examples

The RAF Score Formula

Knowing how to calculate RAF scores is a foundational skill for risk adjustment teams. The Risk Adjustment Factor (RAF) score is calculated using a formula that combines demographic characteristics, clinical conditions, and condition interactions into a single predictive value. Under CMS-HCC V28, the formula is:

RAF Score = (Demographic Baseline + Sum of HCC Coefficients + Sum of Disease Interaction Factors) ÷ Normalization Factor

• Demographic Baseline: A starting value based on age, sex, dual-eligibility status, and living setting that represents the predicted cost of a member before any clinical conditions are considered
• HCC Coefficients: CMS-published risk weights assigned to each qualifying Hierarchical Condition Category after hierarchy rules are applied
• Disease Interaction Factors: Additional risk weights triggered when specific condition combinations are present, reflecting the compounding cost impact of comorbidities
• Normalization Factor: A CMS-applied divisor that maintains budget neutrality across the Medicare Advantage program

Each component introduces specific rules and data requirements. The sections below walk through each step with the precision needed to reproduce CMS calculations.

Step 1: Demographic Baseline

Every RAF score calculation begins with the demographic baseline — the predicted cost of a member based solely on who they are, before any clinical conditions are added. CMS publishes demographic coefficients in the annual rate announcement.

• Age Bands: CMS groups members into 5-year age bands (0-34, 35-44, 45-49, 50-54, 55-59, 60-64, 65-69, 70-74, 75-79, 80-84, 85-89, 90-94, 95+). Each band carries a different baseline coefficient reflecting age-related cost patterns. Older age bands generally receive higher baselines.
• Sex: Male and female members receive different baseline coefficients within each age band, reflecting documented differences in healthcare utilization patterns and costs.
• Dual-Eligibility Status: Members eligible for both Medicare and Medicaid (dual-eligible) receive adjusted baselines. Full dual-eligible, partial dual-eligible, and non-dual categories each carry different coefficients, with full dual-eligible members receiving the highest demographic adjustments.
• Living Setting: The model distinguishes between community-dwelling members, community low-income members, and institutionalized members (long-term care facility residents). Institutional baselines are substantially higher than community baselines.
• Example Baseline: A 72-year-old male, non-dual-eligible, community-dwelling member receives a demographic baseline of approximately 0.367 under V28. A 72-year-old female with the same characteristics receives approximately 0.329. These values shift each payment year as CMS updates coefficients.

The demographic baseline is the floor of the RAF score. A member with no documented HCCs receives only their demographic value — making it the starting point from which clinical complexity builds the score upward.

Step 2: ICD-10 to HCC Mapping

The next step converts clinical diagnoses into the condition categories that carry risk weights. This mapping is where coding accuracy directly impacts the RAF score.

• ICD-10-CM Codes: Diagnosis codes from qualifying face-to-face encounters during the collection year are the input. Only codes from acceptable provider types (MD, DO, PA, NP) and acceptable encounter settings qualify for risk adjustment.
• CMS Crosswalk: CMS publishes an annual ICD-10-to-HCC crosswalk that maps each qualifying diagnosis code to its corresponding HCC. Under V28, approximately 7,770 ICD-10 codes map to 115 HCCs. Codes not in the crosswalk do not contribute to the RAF score.
• Many-to-One Mapping: Multiple ICD-10 codes can map to the same HCC. For example, E11.21 (Type 2 diabetes with diabetic nephropathy), E11.22 (Type 2 diabetes with diabetic chronic kidney disease), and E11.29 (Type 2 diabetes with other diabetic kidney complication) all map to the same diabetes HCC under V28.
• Non-Mapping Codes: Not all ICD-10 codes are risk-adjustable. Codes for acute conditions with low predicted ongoing cost, symptom codes, and codes for conditions with insufficient cost data do not map to HCCs. Under V28, 2,294 previously mapped codes were removed from the crosswalk.
• V28 Mapping Changes: The transition from V24 to V28 fundamentally changed the crosswalk. Organizations must validate their coding against the current V28 mappings — codes that mapped to HCCs under V24 may no longer do so, and vice versa.

Step 3: Hierarchy and Coefficients

After mapping ICD-10 codes to HCCs, CMS applies hierarchy rules that prevent double-counting within disease families, then assigns risk coefficients to remaining HCCs.

• Disease Families: V28 organizes 115 HCCs into 26 disease families (e.g., Diabetes, Heart Failure, Chronic Kidney Disease, Dementia). Within each family, HCCs are arranged from most severe to least severe.
• Hierarchy Application: When a member has multiple HCCs within the same disease family, only the highest-severity HCC is retained. If a patient has both HCC 37 (Diabetes with Acute Complication) and HCC 38 (Diabetes with Chronic Complication), the hierarchy keeps only HCC 37.
• Constraining (V28 Change): The most significant V28 innovation. Under constraining, related HCCs within the same disease family carry identical coefficients regardless of severity. All diabetes HCCs (except pancreas transplant status) share the same weight of approximately 0.166. This eliminated the financial incentive to upcode within a disease family.
• Coefficient Assignment: Each surviving HCC receives its CMS-published coefficient. These values are derived from regression analysis of actual Medicare claims data and represent the predicted incremental cost of each condition category. Coefficients range from approximately 0.06 for minor conditions to over 2.0 for the most complex conditions.
• Cross-Family Accumulation: HCCs from different disease families accumulate additively. A patient with diabetes (one family), CHF (another family), and COPD (another family) receives the sum of all three HCC coefficients in addition to their demographic baseline.

Understanding how risk adjustment applies hierarchies is critical for predicting which documentation will actually impact the RAF score and which conditions are superseded by higher-severity diagnoses.

Step 4: Disease Interactions

The final clinical component of the RAF score formula accounts for the fact that certain disease combinations produce costs greater than the sum of their individual HCC weights.

• Interaction Terms: CMS defines specific condition pairs or groups that trigger additional risk weights when both conditions are present. These interactions are empirically derived from claims analysis showing compounding cost patterns.
• Diabetes + CHF Interaction: One of the most common interaction terms. When a member has both a qualifying diabetes HCC and a qualifying heart failure HCC, an additional coefficient is added to the RAF score beyond the individual condition weights.
• CHF + COPD Interaction: The combination of congestive heart failure and chronic obstructive pulmonary disease triggers an interaction factor reflecting the clinical complexity and cost amplification of managing both cardiopulmonary conditions simultaneously.
• Disabled/Disease Interactions: Certain interactions apply specifically to disabled (under-65) populations, where the cost impact of condition combinations differs from the aged population.
• V28 Updates: V28 revised interaction terms to reflect current clinical cost patterns and the new HCC structure. Some V24 interactions were removed, modified, or replaced with new interaction definitions aligned to V28 disease families.
• Documentation Implication: Disease interactions reward comprehensive documentation. A provider who documents only diabetes but fails to document the coexisting CHF loses not just the CHF HCC coefficient but also the interaction factor — a double revenue impact.

Worked Example With Real Numbers

Consider a 74-year-old female Medicare Advantage member, non-dual-eligible, community-dwelling, with the following documented conditions from qualifying encounters:

• Documented Conditions: Type 2 diabetes with chronic kidney disease (E11.22), Congestive heart failure, unspecified (I50.9), Major depressive disorder, recurrent, moderate (F33.1), COPD with acute exacerbation (J44.1)
• Step 1 — Demographic Baseline: 74-year-old female, non-dual, community = approximately 0.346
• Step 2 — HCC Mapping: E11.22 maps to Diabetes HCC (approximately 0.166 under V28 constraining). I50.9 maps to Heart Failure HCC (approximately 0.368). F33.1 maps to Specified Major Depressive Disorder HCC (approximately 0.309). J44.1 maps to COPD HCC (approximately 0.280).
• Step 3 — Hierarchy: No hierarchy conflicts in this case — each condition falls in a separate disease family. All four HCCs survive. Total HCC coefficients = 0.166 + 0.368 + 0.309 + 0.280 = 1.123
• Step 4 — Interactions: Diabetes + CHF interaction = approximately 0.121. CHF + COPD interaction = approximately 0.098. Total interaction factors = 0.219
• Raw Score: 0.346 (demographic) + 1.123 (HCCs) + 0.219 (interactions) = 1.688
• Normalized Score: 1.688 ÷ normalization factor (approximately 1.015 for 2026) = approximately 1.663

This member's RAF score of 1.663 indicates she is expected to cost approximately 66% more than the average Traditional Medicare beneficiary. At a county base rate of $950 PMPM, this generates approximately $1,580 in monthly CMS payment.

Tools for RAF Score Calculation

• Free RAF Score Calculator: The RAF Score Calculator supports up to 50 ICD-10 codes per calculation with V28, V24, ESRD, and RxHCC models. Enter demographics and diagnosis codes to see the step-by-step score breakdown — no registration required.
• RAF Score API: The RAF Score API provides programmatic access to patient-level risk calculations with HCC mapping details, hierarchy application, interaction factors, and care gap analysis. Designed for integration into EHR systems, population health platforms, and risk adjustment workflows.
• Batch Processing: For population-level analysis, RAF Batch Processing handles 100,000+ member records per batch with detailed per-member score breakdowns, HCC-level audit trails, and aggregate risk analytics for pre-audit validation and revenue forecasting.
• V28 Crosswalk Validation: Automated tools that validate ICD-10-to-HCC mappings against the current CMS crosswalk catch codes that no longer map under V28 and identify new mapping opportunities — preventing both overcoding and missed HCCs.