Two Points Short: What Alignment's Star Ratings Suit Says About How Stars Works

The short version

CMS released the 2027 Medicare Advantage Star Ratings on October 8, 2026, and Alignment Healthcare sued the next day. Its largest contract, H3815, holds 75% of its members and fell from 4.0 to 3.5 stars, below the line where quality bonus payments begin.

I rebuilt the rating from CMS's public data tables to see what drove the drop. Two Health Outcomes Survey measures, whose weight tripled for 2027, each landed one percentage point below the next star. In my model, removing the ten contested measures would probably lift H3815 back to 4.0, and the margin is thin.

CMS can argue that Alignment had real slippage in medication adherence and clinical measures. Both sides can feel they have a fair case. One way of resolving would be to settle measure specifications and weights in regulation before the performance year starts, soften the payment cliff at 4.0, and move outcome measurement onto a more solid evidence footing.

What happened

Alignment filed suit in the U.S. District Court for the District of Columbia on October 9, 2026. According to Healthcare Dive and Fierce Healthcare, the complaint makes four claims:

• CMS set key measure details through its annual Technical Notes instead of notice-and-comment rule-making.
• CMS again used ten measures a federal court had already found invalid.
• The share of Alignment members in plans rated 4 stars or higher fell to 25%.
• More than $200 million in 2028 payments is at stake.

Alignment asked the court to vacate the 2027 ratings and order a recalculation before the 2028 selling season. Healthcare Dive reports that Alignment asked CMS for that recalculation on October 6 and was refused on October 7. The complaint itself was not publicly posted when I wrote this, so the details above come from press accounts.

The ten measures come from Clover Insurance v. HHS. On May 27, 2026, the Southern District of Georgia held two sets of ten measures unlawful as applied to Clover:

• First set: outside the data sources the statute allows.
• Second set: never adopted "by regulation" as the Medicare statute requires for rules governing payment.

CMS's June 17 memo offered all plans an optional, higher-of recalculation of their 2027 quality bonus ratings under the first holding only. It appealed the ruling to the Eleventh Circuit on July 21.

Alignment is already in court over the second set. Its 2026-ratings case before Judge Christopher Cooper (No. 1:26-cv-02441) argues that removing those measures lifts H3815, H4961 and H3443 from 4.0 to 4.5 for 2026. That would raise its rebate share from 65% to 70% of the gap between bid and benchmark. The amended complaint puts the value at more than $50 million; Alignment's summary-judgment brief puts it at about $125 million. Cross-motions were fully briefed on October 6, and I found no ruling as of October 14.

Alignment's 8-K of October 8 puts H3815 at about 75% of its health plan membership. It attributes the decline to:

• higher industry cut points;
• weaker results on triple-weighted survey and Part D measures;
• the removal of the Excellent Health Outcomes for All bonus after the performance period had ended.

The filing says there is no revenue effect in 2026 or 2027 and that the company aims to return H3815 to 4.0 or higher.

One contract carries the company

CMS's September 2026 enrollment file counts 298,618 Alignment members across eight contracts, and 224,262 of them (75.1%) are in H3815. Every other rated contract kept 4 stars or better for 2027.

The 2026 column reflects CMS's July 2026 update after the Clover recalculation. Ratings are from the CMS Star Ratings data tables; enrollment is from the CMS Monthly Enrollment by Contract report.

This table settles the two figures in the coverage, 98% and 100%:

• 2026: every rated contract was at 4 stars or higher. That is 100% of members in rated contracts and 98.1% of all members.
• 2027: the contracts at 4 or higher hold 73,491 members, or 24.6%, which matches the 25% in the complaint.

CMS's 2027 fact sheet puts about 71% of MA-PD enrollees in contracts rated 4 stars or higher for 2027, up from about 67% for 2026.

How the score works, and what changed for 2027

An MA-PD contract is rated on up to 43 unique measures, each scored 1 to 5 stars against cut points CMS sets from that year's results. Per the 2027 Technical Notes, the overall rating is built in four steps:

1. Weighted average. Process measures count once, outcome and intermediate-outcome measures three times, patient experience and access twice, and the two improvement measures five times.
2. Reward factor. Up to 0.4 stars, added for contracts whose average is high and whose stars are consistent across measures.
3. Categorical Adjustment Index (CAI). A small adjustment for the share of low-income and disabled members.
4. Rounding. The total is rounded to the nearest half star, so 3.75 becomes 4.0 and 3.74 stays 3.5.

The rounding matters because payment steps at 4.0. Contracts at 4 stars or higher get a 5% bonus on the benchmark they bid against. Rebates are 65% of the bid-to-benchmark gap at 3.5 and 4.0 stars, and 70% at 4.5 and 5.0.

There is also a rule that protects strong plans. If a contract scores 4 stars or higher without the two improvement measures, and including them would lower its rating, they are left out. That is why H5296 shows 3.5 for Part C, 3.5 for Part D and 4.5 overall.

The biggest change for 2027 is that two Health Outcomes Survey (HOS) measures went from weight 1 to weight 3: "Improving or Maintaining Physical Health" (C04) and "Improving or Maintaining Mental Health" (C05). CMS's fact sheet says this followed a substantive change to their specifications and was adopted through earlier notice-and-comment rule-making.

Other changes for 2027:

• Added, at weight 1: two new drug-safety measures (opioid-benzodiazepine overlap and anticholinergic polypharmacy), a respecified functional status assessment and a respecified colorectal screening measure.
• Retired: pain assessment and medication reconciliation.
• Moved to display only: medication therapy management completion.

The ten contested measures in the 2027 numbering are:

• Survey: flu vaccine (C03), physical health (C04), mental health (C05), reducing the risk of falling (C15) and bladder control (C16).
• Patient experience: getting needed care (C21), getting appointments quickly (C22), customer service (C23), rating of health care quality (C24) and care coordination (C26).

For H3815 they carry 15 of its 79 weight points; two of the experience measures had too few responses to score.

Alignment's legal argument targets how these measures are specified in the Technical Notes. That is separate from whether CMS was entitled to raise the weights.

Inside H3815's 3.5

Comparing H3815's measure stars for 2026 and 2027 shows two separate problems: a collapse on one survey measure, and a broad slide on measures nobody is contesting.

The physical health measure carries most of the survey story. H3815 scored 79% in 2026, worth 5 stars, and 66% in 2027, worth 1 star. The cut points barely moved: 2 stars needed 66% in 2026 and 67% in 2027, and 5 stars needed 75% both years. The mental health measure slipped from 82% to 80%, just under the 81% needed for 2 stars. Each measure missed the next star by one percentage point, at triple weight.

Who is in the survey matters. The 2027 HOS results follow members surveyed in 2023 and again in 2025 who were still enrolled at follow-up, with case-mix adjustment. Alignment's 8-K says about half its members joined in the last two years. Those members are not in this cohort, so rapid growth does not explain the result; it is a two-year look back at care for long-tenured members.

The uncontested slide is just as large:

• Medication adherence: diabetes, hypertension and statin adherence fell from 91%, 93% and 92% to 89%, 89% and 88%. Each dropped to 3 stars, at weight 3.
• Clinical measures: colorectal screening, osteoporosis management, diabetes eye exams and blood pressure control each lost a star. SNP care management fell from 92% to 79%.
• Gains: patient experience improved, appeals review recovered, and the drug plan improvement measure rose a star.

Disasters do not enter into it. 21% of H3815's members lived in 2025 disaster areas, below the 25% threshold CMS uses for adjustments.

Would winning restore 4 stars?

Probably, by a narrow margin. I rebuilt CMS's overall rating calculation from the public data tables and Technical Notes, then changed one thing at a time for H3815. The model is my reconstruction, not CMS's own code; how closely it matches is covered at the end.

Every path that softens the two survey outcome measures gets H3815 to 4.0:

• Removing the ten contested measures lifts it from 3.40 to 3.87.
• Removing only the two survey measures, or returning them to weight 1, also gets there.
• So would scoring one percentage point higher on each. Two points of survey results were worth a half star.

The weight change hits twice. Two 1-star measures at weight 3 pull the average down, and they also make the contract's stars uneven. The reward factor pays for a high average and consistency, so at 3.40 H3815 earns none. In each scenario above it earns 0.2, and that is what carries it past 3.75.

That is also where the uncertainty sits. CMS told the court that any recalculation would remove the measures for every contract and re-derive the reward-factor thresholds. When I approximate that, H3815 still reaches 4.0, but it clears the reward-factor bar by only about 0.04 of a star. A court win is likely, though not certain, to restore 4.0 for 2027.

Alignment's smaller contracts are steadier. Removing the ten measures would move H3443 from 4.0 to 4.5, and H4961 stays at 4.5.

An aside: be careful what you sue for

The briefs in the 2026 case contain an exchange worth noting.

• CMS's argument. All Star Ratings measures went through rule-making, but none of their detailed specifications are written into regulation; the 2018 rule said CMS would not codify them. If Alignment's theory is right, every measure shares the defect, and CMS asked the court for room on remand to "consider whether" all of them should be excluded (reply, October 6, 2026).
• Alignment's reply. Its September 22 brief concedes that other measures "could be subject to the same as-applied attack" and argues it only has standing to challenge the ones that hurt it.

CMS is unlikely to follow through. According to Alignment's brief, when Clover won on the same ten measures, CMS removed just those ten. A rating built on no measures would not satisfy the statute either. Even in my model, removing H3815's six other 5-star measures on top of the ten contested ones still leaves it at 4.0.

The exchange is a reminder that the legal theory reaches well past ten measures. A win pushes the whole program toward specifications fixed in regulation, which is slower to change for CMS and for every plan.

The case for CMS

Stars is doing what it was designed to do: steer money toward outcomes.

• It rewards results over process. Outcomes, intermediate outcomes and improvement carry three to five times the weight of process measures. Raising the survey outcome measures to weight 3 continues that direction, and CMS adopted the change through rule-making.
• It keeps pruning. Measures that stopped telling plans apart, such as pain assessment and medication reconciliation, were retired for 2027.
• Annual specifications give it flexibility. CMS updates them each year in the Technical Notes. That is how it applied disaster adjustments for the California wildfires and Texas storms. Its August brief notes that case-mix coefficients come from each year's data and cannot be written into regulation in advance.
• Relative cut points protect the program. A fixed bar would let every plan drift to 4 stars, and the bonus would become a flat subsidy. Rewarding performance against peers keeps the 4-star label meaningful to the beneficiaries who choose plans by it.

The system is also not crushing the industry. About 71% of MA-PD enrollees are in contracts rated 4 stars or higher for 2027, up from about 67% for 2026. CMS has to manage a quality bonus program that MedPAC has estimated adds billions of dollars a year to Medicare spending. Every plan that wins a recalculation in court raises the cost of the program for everyone else.

The case for plans

Plans are asked to invest on long horizons against rules they see late.

• The timeline is long and the final rules arrive at the end of it. The 2027 ratings rest mostly on care delivered in 2025, a survey cohort that began in 2023, and Technical Notes finalized on September 30, 2026. They pay out in 2028.
• The scoring has three edges. Cut points move with the industry, a half-star decides the bonus, and the reward factor adds another threshold on top. A plan can improve and still fall.
• The weight sits on a noisy measure. Tripling the weight of a survey measure with sampling noise and a two-year lag concentrates a lot of money on a sample survey. H3815 lost a half star on two measures that each missed by one percentage point.
• Rules change after the fact. Alignment's 8-K says CMS retroactively eliminated the Excellent Health Outcomes for All bonus after the performance period had ended, which no plan could have planned for.
• Recourse is mostly the courts. In both the July and October episodes, Alignment says CMS refused its recalculation request within a day. The courts are the remaining path, and that is a slow way to settle a rating.

Concentration amplifies all of this. With 75% of its members in one contract, a half star on H3815 moves most of the company.

The feature in the friction

The litigation is slow and expensive, and it is doing something useful. Each case puts a piece of the method on the record:

• the specifications behind each measure;
• the reasoning for each weight;
• the exact recalculation steps CMS would follow.

CMS's own declarations in the 2026 case now explain, step by step, how a recalculation would treat improvement measures, the reward factor and the CAI. That is more transparency than the Technical Notes alone have offered.

The pressure is also sorting what has to be fixed in advance from what can be tuned each year. Measure specifications and weights shape how plans invest, so they belong in regulation before the performance year. Cut points, case-mix coefficients and disaster adjustments depend on each year's data, so they should stay annual, calculated by methods that are already set in regulation. Measures that survive this scrutiny will be harder to challenge and more stable. CMS gets firmer legal footing, and plans get a longer runway.

The data also make a point no court can. The fastest route back to 4 stars for H3815 runs through a few survey points and a few points of medication adherence. Falls prevention, mental and physical health support, and adherence programs move those numbers. Litigation can fix the rules, and care delivery is what moves the score. It just takes time.

My humble opinion on how it might work better

Four changes would serve CMS, plans and beneficiaries better than another round of lawsuits.

1. Bridge the litigation period. Extend the June 17 approach, a voluntary higher-of recalculation, to 2028 quality bonus payments while the Eleventh Circuit and the D.C. cases run. It is the cheapest relief for CMS. It gives plans certainty before 2028 bids are due in June 2027, and it keeps beneficiaries' supplemental benefits stable.
2. Lock the rules before the performance year. Put measure specifications and weights into regulation at least a year before measurement begins. The measure stewards revise their specifications every year: NCQA for HEDIS and the Pharmacy Quality Alliance for the drug measures. The practical route is to adopt a dated version by reference in each year's rule, a year ahead, rather than copy every detail into regulation text. Cut points and other data-driven coefficients should stay annual, since they depend on each year's results. Their methods are already in regulation: 42 CFR 422.166 sets out the clustering, outlier removal and 5-point guardrail for cut points, along with the reward factor and the CAI. What is missing is the same discipline for the specifications themselves, published with those formulas so plans know how they will be scored before the year starts.
3. Soften the cliff at 4.0. Phase the bonus in across a band around 4.0 instead of a single step. MedPAC's Medicare Advantage Value Incentive Program proposal already uses continuous scoring. A version that holds total spending flat would remove most of the incentive to litigate a half star.
4. Measure outcomes on better evidence. Survey measures with small samples and a two-year lag are the wrong place to concentrate weight. As interoperability matures, shift outcome measurement toward claims and clinical data with larger populations. Until then, publish confidence intervals for survey measures and avoid tripling the weight of any measure whose result depends on a few points of survey response.

Basically give plans a chance to invest in multi-year care programs that are aligned to the outcomes that CMS wants to see and allow the bonuses to follow those programs.

What to watch

About the model

The scenario results are my reconstruction of CMS's calculation from public files. They are not CMS's numbers, and they should be read as informed estimates.

What it does. It follows the 2027 Technical Notes (dated September 30, 2026):

• Attachment G measure weights;
• the weighted average with and without improvement measures, and the improvement-measure protection for 4-star contracts;
• the reward factor, using CMS's published thresholds and weighted-variance formula;
• the CAI values from the 2027 CAI supplement;
• the hold harmless for new measures at contracts with 25% or more of members in disaster areas;
• CMS's rounding table.

Measure stars come from the CMS 2027 Star Ratings data tables released October 8, 2026.

How well it matches. It reproduces CMS's published 2027 overall rating exactly for 426 of 496 MA-PD contracts (86%) and within a half star for all 496. It matches all seven rated Alignment contracts exactly.

What it leaves out:

• It does not recompute improvement-measure stars or the CAI with measures removed, which CMS would.
• It does not model consolidations or data-integrity reductions.
• Where a scenario lands within a few hundredths of a rounding or reward-factor threshold, the outcome could go either way. Removing the ten contested measures under CMS's stated recalculation method, which I approximate by shifting the published reward-factor thresholds, is one such case.

I did not read the October 9 complaint, which was not publicly available when this was written. References to it rely on press reports.

References

CMS

1. CMS, 2027 Medicare Advantage and Part D Star Ratings fact sheet, October 8, 2026.
2. CMS, 2027 Part C and D Star Ratings Technical Notes, last updated September 30, 2026.
3. CMS, 2027 Star Ratings data tables and 2026 Star Ratings data tables, via Part C and D Performance Data.
4. CMS, 2027 Categorical Adjustment Index measure supplement.
5. CMS, Monthly Enrollment by Contract, September 2026.

Court filings (all from the Georgetown litigation tracker)

6. Alignment Healthcare v. HHS, amended complaint, No. 1:26-cv-02441-CRC (D.D.C.), July 31, 2026.
7. Defendants' response and cross-motion for summary judgment, August 28, 2026.
8. Plaintiffs' response and reply, September 22, 2026.
9. Defendants' reply, with second Goldstein declaration, October 6, 2026.

Company filings and analysis

10. Alignment Healthcare, Form 8-K, October 8, 2026.
11. Crowell & Moring, Clover Insurance v. HHS: S.D. of Georgia holds 20 Star Ratings measures unlawful.
12. Avalere Health, Post-Clover and 2027 quality bonus payment recalculation.
13. MedPAC, Medicare Advantage Value Incentive Program, April 2020.

News coverage

14. Healthcare Dive, Alignment sues CMS over 2027 Medicare Advantage star ratings crash, October 13, 2026.
15. Fierce Healthcare, Alignment Healthcare sues CMS over its 2027 Medicare Advantage star ratings, October 12, 2026.
16. Health Exec, Insurer files lawsuit as CMS releases 2027 Medicare Advantage star ratings.

← All musings · /healthcare