# Your risk adjustment scores move every year because the model was recalibrated, not because patients changed.

![A risk adjustment lead comparing model version records against year on year score movement on a health plan office screen, used to show where score changes are actually explained.](https://neurobird.com/riskadjustment/riskadjustment-hero.webp)
*A score without its model version is a number without a scale.*

> Risk adjustment HCC coding is the practice of capturing patient diagnoses accurately enough that a health plan is paid for the risk it actually carries. Diagnoses map into hierarchical condition categories, those categories feed a regression model, and the model produces a risk score. The model is not fixed. It is recalibrated on new data, versioned, and normalised so that the average score stays at a constant, which means a score can move without a patient changing at all.

- URL: https://neurobird.com/riskadjustment/
- Product: Neurobird Risk Adjustment HCC Coding Operations Platform
- Niche: risk adjustment HCC coding
- Buyer: health plans and provider groups
- Status: in development, open for early access
- Updated: 2026-08-26

## What Neurobird Risk Adjustment HCC Coding does

- Record which model version and benefit year a given score was produced under.
- Link each captured condition back to the documentation that supports it, so a category can be defended.
- Separate coding driven movement from recalibration driven movement when scores change.
- Track Part C and Part D models distinctly, because they are related but not the same.

## How it works

1. **Capture** Conditions are recorded against the documentation supporting them, so the category has evidence behind it rather than an assumption.
2. **Submit** Submissions are tracked against the processing systems that acknowledge them, so a diagnosis that never landed is visible rather than presumed.
3. **Explain** Score movement is attributed between coding, population and model change, which is the question leadership actually asks.

## From the source material

> HCC Groups refers to HCCs that are grouped together in the HHS-HCC risk adjustment model.

Source: CMS, potential updates to HHS-HCC risk adjustment models, https://www.cms.gov/cciio/resources/regulations-and-guidance/downloads/potential-updates-to-hhs-hccs-hhs-operated-risk-adjustment-program.pdf

## Industry context

- **1.0** Recalibrated models retain an average 1.0 risk score, while individual and plan average scores may still move. (source: CMS Medicare Managed Care Manual chapter 7, https://www.cms.gov/regulations-and-guidance/guidance/manuals/downloads/mc86c07.pdf)
- **70.5** CMS guidance sets out the similarities and differences between the Part D drug model and the Part C medical model in a dedicated section. (source: CMS Medicare Managed Care Manual chapter 7, https://www.cms.gov/regulations-and-guidance/guidance/manuals/downloads/mc86c07.pdf)
- **V06a** Candidate model updates are presented by comparing regression results for the current version against the proposed V06a models. (source: CMS potential updates to HHS-HCC models, https://www.cms.gov/cciio/resources/regulations-and-guidance/downloads/potential-updates-to-hhs-hccs-hhs-operated-risk-adjustment-program.pdf)
- **2018** From the 2018 Payment Notice, masked enrollee level data has been collected and used to recalibrate the models and update the methodology. (source: CMS potential updates to HHS-HCC models, https://www.cms.gov/cciio/resources/regulations-and-guidance/downloads/potential-updates-to-hhs-hccs-hhs-operated-risk-adjustment-program.pdf)

## Pricing

- Single plan: $180 per month
- Multi line: $460 per month
- Enterprise: $920 per month

## Questions

### What is HCC risk adjustment coding?

It is the practice of documenting and coding patient conditions so they map into hierarchical condition categories used to predict expected costs. Those categories, together with demographic information and enrolment duration, feed a regression model that produces a risk score for each person. Payment follows the score. The coding work is therefore not administrative tidying; it is the input to a payment model, and an unrecorded condition is an unfunded one.

### Why do risk scores change without patients changing?

Because the model changes. Models are recalibrated on newer data, and CMS applies a normalisation factor so the average risk score stays at 1.0 after recalibration. Recalibrated models retain that average, but individual beneficiaries' scores may change, and so may plan averages, depending on each person's combination of diagnoses. Movement in an average is therefore not automatically evidence of better or worse coding.

### What is a model version?

It is a specific published version of the classification and its coefficients. CMS considers classification updates and compares candidate models against the current ones, for example presenting regression results for proposed models alongside the version in force. Because categories can be regrouped between versions, comparing two years without recording which version produced each score gives a misleading answer.

### Are the Part C and Part D models the same?

No, though they are closely related. CMS guidance describes the similarities and differences between the Part D drug model and the Part C medical model in its own section. They serve different purposes and behave differently, so analysis that blends them produces confident conclusions about the wrong population.

### Is the software available now?

No. It is in development and open for early access. We are building the model version and normalisation tracking with risk adjustment teams who have to explain score movement to a board.

## Sources

- [CMS, potential updates to HHS-HCC risk adjustment models](https://www.cms.gov/cciio/resources/regulations-and-guidance/downloads/potential-updates-to-hhs-hccs-hhs-operated-risk-adjustment-program.pdf)
- [CMS, Medicare Managed Care Manual chapter 7](https://www.cms.gov/regulations-and-guidance/guidance/manuals/downloads/mc86c07.pdf)

## Contact

- office@neurobird.com
- https://neurobird.com/