Cut points in Medicare Star Ratings

Cut points are the numeric thresholds CMS sets each year that determine how many stars, from 1 to 5, a Star Ratings measure earns. For most measures CMS draws them with a clustering algorithm applied after Tukey outlier deletion and within guardrails. Cut points are set retrospectively, after the measurement year closes, and move from year to year, which is why forecasting where they will land matters more than matching last year's thresholds.

What cut points are

Cut points are the numeric thresholds in the CMS Medicare Star Ratings that turn a measure rate into a star score. Each measure has a set of cut points that divide performance into five bands, so a contract earns 1, 2, 3, 4, or 5 stars on that measure depending on where its rate falls. A measure rate of 84 percent might earn 4 stars one year and 3 the next, entirely because the cut points moved.

The star scores from every measure then feed the weighted average that produces the Part C summary, the Part D summary, and the overall Star Rating. Cut points are the hinge: they are where a raw number becomes a rating.

How CMS sets cut points

For the non-CAHPS measures, CMS does not pick round numbers. It draws the cut points from the distribution of all contracts' performance using a sequence of steps:

• Tukey outlier deletion. CMS first removes statistical outlier contracts using the Tukey outer-fence rule, so a few extreme scores cannot distort the thresholds.
• Mean resampling and clustering. It then runs mean resampling and a hierarchical clustering algorithm on the remaining scores to find the natural breaks that become the five star bands.
• Guardrails. A guardrail limits how far a cut point can move from the prior year, generally no more than 5 percentage points, or 5 percent of the restricted range, so the thresholds cannot lurch in a single year.

Cut points are set retrospectively and move

Two properties make cut points hard to manage. First, they are set retrospectively. CMS draws them from how every contract performed during the measurement year and does not publish them until the Star Ratings are released the following fall. A team doing the work does not yet know the line it has to clear.

Second, the cut points move. They are relative to the field, so when most plans improve on a measure, the thresholds rise and a contract has to improve just to hold its star. The Tukey method, first applied for the 2024 Star Ratings, tends to push the thresholds up further. A contract can post the exact rate it posted last year and still drop a star because the cut point climbed past it.

Why forecasting the cut points matters

Put those two properties together and the operating problem is clear: the target moves every year, and you cannot see it until the year is over. Aiming at last year's cut points is aiming at a line that has probably already shifted up. The teams that hold their ratings are the ones that forecast where the cut points are likely to land under the current methodology and manage each measure to that forecast during the year, while gaps can still be closed.

This is what glide-path forecasting does. It projects the threshold, tracks each measure against the projection, and surfaces which gaps still change the star and which no longer can.

Common mistakes teams make with cut points

• Treating last year's cut points as this year's target. The thresholds move, and under Tukey they usually move up.
• Waiting for the published cut points. By the time CMS releases them, the measurement year is closed and nothing more can be done for that year.
• Ignoring the guardrail's direction. The guardrail caps how far a cut point moves but not which way. A capped move is still a move, and it is often upward.
• Forecasting the rate but not the threshold. Projecting your own measure rate is half the picture. The star depends on where that rate sits against a cut point you also have to project.

How Pelica handles cut points

Pelica's Quality and Stars Copilot runs glide-path forecasting for HEDIS and Star Ratings measures, projecting where cut points are likely to land and closing gaps before CMS sets the thresholds rather than after. On the three triple-weighted Part D adherence measures, customers hold 96 percent medication adherence.

Frequently asked questions

What are cut points in Star Ratings?

Cut points are the numeric thresholds CMS sets each year that determine how many stars, from 1 to 5, a Medicare Star Ratings measure earns. For most clinical and process measures, CMS draws the cut points with a hierarchical clustering algorithm applied after it removes Tukey outliers and within guardrails that limit how far a cut point can move year to year. A contract's measure rate is compared to the cut points to assign that measure its star score.

How does CMS set cut points?

For non-CAHPS measures, CMS removes statistical outliers using the Tukey outer-fence method, then runs mean resampling and hierarchical clustering on the remaining contract scores to find natural breaks in the distribution. Those breaks become the five star bands. Guardrails then limit the change to no more than 5 percentage points, or 5 percent of the restricted range, from the prior year so the thresholds cannot lurch.

Are cut points set before or after the measurement year?

After. Cut points are set retrospectively, based on all contracts' performance during the measurement year, and are not published until the Star Ratings are released the following fall. A contract does not know the exact thresholds while it is still doing the work, which is why forecasting where cut points will land is more useful than reconciling against last year's numbers.

Why do cut points change every year?

Cut points are relative, not fixed. Because they are drawn from the distribution of all contracts' performance in that year, they move as the industry improves or declines. When most plans get better on a measure, the cut points rise and a plan must improve just to hold its star. The Tukey method, introduced for the 2024 Star Ratings, tends to push cut points up further, which is why a stable raw rate can still lose a star.