Five-Star Quality Measures: Interpret Points, Missing Measures and Changing Thresholds

Separate a measure value, its points and its stars

A nursing home's clinical dashboard may contain dozens of measures. The September 2026 Five-Star guide uses a selected set of 15 for the QM rating: nine long-stay and six short-stay measures. Some award up to 150 points; others up to 100. Equal-weight averaging of raw percentages cannot reproduce the rating.

Create a rating worksheet with one line per included measure: specification, collection period, eligible denominator, reported or imputed status, value used for scoring, point-table version and assigned points. Preserve whether higher or lower performance is favorable. Do not assume a shared measure name means two CMS programs use the same population, period or calculation.

The guide generally uses four quarters for QM ratings, while successful return to home and community uses two years. Some measures combine quarterly results with denominator weighting; others are calculated across a full period. Averaging four displayed percentages without their denominators can yield a different number.

Resolve missing data before adding points

The guide's minimum-data rules distinguish public reporting from rating construction. Where at least five of nine long-stay measures have adequate data, CMS can impute the smaller-denominator long-stay measures for scoring. The corresponding short-stay condition is at least four of six. If only one component qualifies, the overall QM rating follows that component. If neither qualifies, no QM rating is generated.

For eligible imputation, the available facility observations are retained and the missing portion is supplemented using the state average to reach the specified minimum. Measures using imputed data are not displayed as individual public values. The guide also identifies a public-reporting minimum of 25 for successful return to home and community. A public blank therefore does not tell an analyst which scoring path CMS used.

In the worksheet, record “not publicly displayed; scoring treatment under review” until the actual denominator and applicable rule are established. Replacing every blank with zero will systematically distort the reconstruction.

A finished component-to-rating calculation

This fictional facility has sufficient data for both components, no applicable audit downgrade and verified point assignments. Its long-stay worksheet sums to 720 points. The short-stay worksheet is:

The maximum unadjusted short-stay score is 800, compared with 1,150 for long-stay. CMS scales short-stay points by 1,150 ÷ 800. Here, 520 × 1,150 ÷ 800 = 747.5. The combined score is 720 + 747.5 = 1,467.5.

Under the July 2026 thresholds in Table 5, the long-stay component is four stars, the adjusted short-stay component is five stars, and the combined QM score is four stars. Averaging four and five stars would not reproduce this calculation. The overall facility Five-Star rating is another calculation, anchored in health inspections and modified by staffing and QM ratings; it is not the combined QM score.

Explain specification and threshold changes separately

CMS raised aggregate QM rating thresholds in July 2026 while stating that individual measure point thresholds did not change in that update. A stable point total can therefore occupy a different aggregate rating band. Retain both the old and new threshold versions when explaining a change rather than attributing every star movement to resident outcomes.

The guide also records substantive measure changes. The January 2026 long-stay antipsychotic measure began incorporating claims and encounter data alongside MDS data; the April update corrected specified schizophrenia exclusions and reset its scoring cut points. Those changes require a different explanation from an aggregate threshold adjustment. A local MDS-only extract is not a complete substitute for the respecified measure.

Make the management conclusion specific

For the example, the defensible conclusion is: “Both components qualify; verified points produce a four-star overall QM rating under the July 2026 thresholds, despite a five-star short-stay component.” The evidence does not establish a particular facility-wide overall rating or prove the quality of every service.

When evaluating a platform, require the report to preserve measure periods, scoring versions and missing-data decisions. A useful explanation should survive a refresh and show which change came from actual source data, calculation method or rating thresholds. That lets the quality team choose a clinical improvement project based on the underlying issue rather than chasing a star movement it has misinterpreted.