Medicare Measure Atlas
Built from the corrected pipeline guide, formulas doc, and measure spec package

How measures become Stars.

This page is the visual teaching layer for Medicare Stars. Start with one measure, learn how its raw score and Star work, then zoom out to see contract applicability, roll-up logic, QBP impact, and the operational exceptions that usually trip teams up.

Choose a measure first Open the roll-up flow Open the QBP bridge Open the pipeline guide Open the formulas doc

Choose a Measure

Use this as the A-to-Z learning path: pick one active 2026 measure, choose a contract type, and walk from raw score to Star to roll-up to business impact without leaving the page.

Measure-first explorer

Business-first by default, technical detail on demand

The explorer uses the expanded 2026 measure workbook as its source of truth and hides the deeper implementation detail until you ask for it. That way business, product, and engineering can all start with the same measure story.

Open the package reference Open the usage guide Open QBP 101
Audience mode
Keep business-first, reveal technical detail only when needed.
Part filter
Contract type
Family filters
Domain filters
Selected measure

Loading selected measure...

1. At a glance

What it is

2. Contract fit

Does it apply to my contract?

3. Raw score

How the raw score is built

4. Star assignment

How it becomes a Star

5. Roll-up

Where it rolls up

7. Critical exceptions

What can change or suppress it

8. Source trace

Where this story came from

The 101 layer

These are the core words teams mix together. Learn them in the right order and the rest of the pipeline becomes much easier to reason about.

Definition

Raw score

The native result of a measure family: rate, percent, survey score, adjusted outcome, ratio, or index. It is not the same thing as a Star.

Definition

Measure star

The 1-to-5 output assigned after the raw score goes through clustering, CAHPS rules, or improvement-measure rules.

Definition

Domain rating

An unweighted average of measure stars inside one domain, produced only when the domain minimum is met.

Definition

Part C / Part D summary

A weighted average of included measure stars for one Part. It is not a weighted average of domains.

Definition

Overall rating

The MA-PD roll-up across unique Part C and Part D measure stars, with shared measures counted once.

Definition

Reward factor

An adjustment rewarding contracts with stronger performance and lower variance before final rounding.

Definition

CAI

The Categorical Adjustment Index. It adds to or subtracts from interim summary or overall scores based on LIS/DE and disability groupings.

Definition

Half-star rounding

The last step for summary and overall outputs. Do not round early. Domain and summary/overall ratings do not follow the same rounding rule.

Contract type changes the measure universe

Before anyone talks about domains, summaries, or overall stars, the contract type decides which measure families even belong in scope.

Broadest operating footprint

Coordinated Care Plan without SNP

Reports
What drops out
Why it matters in the roll-up

The roll-up ribbon

This is the critical teaching path. The first stages explain the measure stack. The last stages explain how contracts become final stars.

Weighting: no

Contract-type measure gate

Teaching formula
What this stage is really doing

Common implementation mistake

The three truth cards

These three cards solve the biggest conceptual mistake in Stars education: collapsing domain, summary, and overall into one generic roll-up idea.

Weighting: no

Domain rating

Average of rated measure stars in one domain.

Domain_raw = mean(star_i in domain)
  • Requires the domain minimum for that contract type.
  • Used for display, analysis, and Plan Finder domain messaging.
  • Reported domain values are rounded to the nearest integer.
Weighting: yes

Part C / Part D summary

Weighted average of included measure stars for one Part.

Summary_raw = sum(weight_i * star_i) / sum(weight_i)
  • Uses summary minimum rules, not domain minimum rules.
  • Improvement measures do not count toward the minimum.
  • Do not replace this with a weighted average of domain ratings.
Weighting: yes

Overall rating

Weighted average of unique Part C and Part D measure stars for MA-PD.

Overall_raw = sum(weight_i * unique_star_i) / sum(weight_i)
  • Only for MA-PD contracts with both summaries.
  • Shared complaint and disenrollment measures count once.
  • Reward factor and CAI happen after this weighted mean.

What Stars affect beyond display ratings

This is the business bridge most teams still miss. Stars are not only a quality label. For Medicare Advantage, they also influence QBP status, benchmark behavior, rebate class, and some contract-lifecycle treatment.

Warning

Payment bridge, not a full ratebook calculator

This section teaches the logic chain from Stars to payment impact. It does not calculate county benchmarks, qualifying-county status, risk-adjusted benchmark amounts, or final rebate dollars.

Open the QBP 101 doc Open the governance doc
Business bridge

Stars result

Teaching formula
What this stage is really doing

Common business misunderstanding

Display output

Public-facing outcomes

Consumers see summary ratings, overall ratings, and sometimes icons. Those display outputs matter, but they are not the whole business story.

  • High performing icon ties to the highest relevant rating.
  • Low performing icon is a multi-year summary-rating rule.
  • Display logic and payment logic overlap, but are not identical.
Payment output

QBP and rebate bridge

For MA plans, Star results affect whether a contract gets a QBP percentage and which rebate class applies once the benchmark exceeds the bid.

  • 4.0+ stars unlock the standard QBP path.
  • Qualifying counties can double the QBP percentage.
  • Rebate class also changes by Star bucket.
Operations output

Lifecycle and governance

Some of the biggest production drifts come from special cases, not from ordinary formulas.

  • Consolidations use special surviving-contract logic.
  • Biased, erroneous, or NR data can override a valid score to 1 star.
  • Low-enrollment and disaster rules can change what counts at all.

Critical gates that change the output

These are not polish rules. They change whether a contract gets a rating and what that rating becomes.

Gate 1

Domain minimum rule

Domain ratings require a domain-specific minimum count.

  • Odd domain total -> divide by 2 and round up.
  • Even domain total -> divide by 2 and add 1.
  • Example: 6 measures in a domain -> minimum 4 rated measures.
Gate 2

Summary minimum rule

Summary ratings use a different minimum rule than domain ratings.

  • Odd required total -> divide by 2 and round to a whole number.
  • Even required total -> divide by 2.
  • Example: 30 required measures -> minimum 15 rated measures.
Gate 3

Improvement measures

C30 and D04 matter, but they are special.

  • They can contribute once rated.
  • They do not count toward the summary minimum.
Gate 4

Shared measures in Overall

Complaint and disenrollment measures span both Parts.

  • Overall includes only one instance of each shared measure.
  • This keeps overall from double-counting the same signal.
Gate 5

Adjustment order

Interim weighted score is not the final displayed star.

  • Reward factor first.
  • CAI second.
  • Half-star rounding last.
Gate 6

Contract applicability

Not every plan reports every measure family.

  • PDPs do not report Part C measures.
  • I-SNP-only contracts have a reduced survey footprint.
  • MSA contracts do not report Part D measures.

Measure family atlas

Once the roll-up model is clear, this is where you learn how raw scores are actually built inside each family.

HEDIS rate

Formula shape
Common modeling error

Implementation posture

    Myth versus fact

    These are the misunderstandings that create the biggest implementation drift between business review, spreadsheets, and code.

    Myth

    Domain-weighted summary

    Myth: Summary and overall are weighted averages of domain ratings.

    Fact: Summary and overall are weighted averages of included measure stars.

    Myth

    One minimum rule

    Myth: The same minimum-measure rule applies everywhere.

    Fact: Domain minimum rules and summary minimum rules are different.

    Myth

    CAI replaces reward

    Myth: CAI replaces reward factor.

    Fact: Reward factor and CAI are separate adjustments with a defined order.

    Myth

    One family fits all

    Myth: Every survey measure is a composite and every pharmacy measure is adherence.

    Fact: Families matter. C25/C26 and D11/D12 prove it quickly.

    Worked walkthrough

    This example is illustrative. It teaches where each number belongs in the pipeline and where weighting enters, without pretending this page is a live calculator.

    Illustrative example

    Contract gate

    What to notice

    The look-alike traps

    These are the places where business language sounds similar, but the formula family underneath is not the same.

    Trap 1

    C25 / C26 versus C27

    • C25 and C26 are single rating questions.
    • C27 is a multi-item composite.
    • All are CAHPS, but not all share the same raw-score structure.
    Trap 2

    C31 / C32 / C33

    • C31 is timeliness.
    • C32 is upheld-decision percentage.
    • C33 is call-center interpreter and TTY availability.
    Trap 3

    D08-D10 versus D11 / D12

    • D08-D10 use PDC math.
    • D11 is MTM completion.
    • D12 is SUPD with IPSD and statin-fill logic.

    The C23 teaching lens

    A score like 81.4 is a transformed adjusted mean, not a plain success percentage. That single idea is still the fastest way to move someone from beginner thinking to real Stars-model thinking.

    81.4 = ((X - 1) / 3) * 100
    X = (81.4 / 100) * 3 + 1 = 3.442