UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA

ALIGNMENT HEALTHCARE, INC.,

Plaintiff,

v. Case No. 25-cv-74 (CRC)

U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES, et al.,

Defendants.

OPINION

This is a dispute over star ratings issued by the Centers for Medicare and Medicaid

Services (“CMS”) to three Medicare Advantage plans offered by Plaintiff Alignment Healthcare,

Inc. Alignment challenges the star ratings on a variety of grounds under the Administrative

Procedure Act and has sought summary judgment on an expedited timeline. The Court will

largely deny its motion for summary judgment and grant summary judgment to the government.

The Court will, however, vacate Alignment’s 2025 star ratings as to plan H3443 because CMS’s

inclusion of two appeals in those ratings was arbitrary and capricious.

I. Background

Given the expedited timeline for this decision, the Court writes primarily for the parties

and presumes that other readers are generally familiar with the relevant legal landscape. The

Court thus provides only a cursory overview of the background for this case.

The Medicare Act establishes a five-star rating system for Medicare Advantage plans on

a scale of one to five stars. See 42 U.S.C. § 1395w-23(o)(4)(A). Alignment Healthcare provides

six Medicare Advantage plans and received its star ratings for 2025 last fall. It now challenges

the ratings issued to three of its plans, claiming that CMS’s methodology for calculating these ratings was flawed or underexplained. Alignment also asked, with the government’s consent, to

expedite this case so that any changes to its plans’ star ratings could be included in its bid

proposals for the 2026 plan year. See ECF 9. Following an accelerated summary-judgment

briefing schedule, the Court heard argument on the parties’ cross motions on June 3, 2025.

II. Standard of Review

At summary judgment, the Court must determine whether the challenged agency action

complies with the APA and is supported by the administrative record. Richards v. INS,

(D.C. Cir. 1977). Under the APA, “[t]he reviewing court shall . . . hold unlawful and

set aside agency action, findings, and conclusions found to be arbitrary, capricious, an abuse of

discretion, or otherwise not in accordance with law[.]”

(2)(A). Arbitrary and

capricious review is “narrow,” Citizens to Preserve Overton Park, Inc. v. Volpe,

(1971), and precludes the Court from “substitut[ing] its judgment for that of the

agency,” Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co.,

(1983). Rather, the Court must determine whether the agency “examine[d] the relevant data

and articulate[d] a satisfactory explanation for its action including a rational connection between

the facts found and the choice made.”

(internal quotation marks omitted). Even if the agency

did not fully explain its decision, the Court may uphold it “if the agency’s path may reasonably

be discerned.” Bowman Transp., Inc. v. Arkansas-Best Freight System, Inc.,

(1974). The Court’s review is limited to the administrative record, Holy Land Found. for Relief

& Dev. v. Ashcroft,

(D.C. Cir. 2003), and the party challenging an agency’s

action bears the burden of proof, City of Olmsted Falls v. FAA,

(D.C. Cir.

2002).

2 III. Analysis

Alignment raises seven objections to CMS’s calculations for its 2025 star ratings: (1)

CMS’s use of the Tukey Outlier Rule is arbitrary and capricious; (2) CMS arbitrarily included

low-reliability enrollee survey data; (3) CMS arbitrarily ignored survey data problems stemming

from Spanish-speaking plan participants allegedly receiving English-language surveys; (4)

CMS’s permitting of oversampling plan enrollees unfairly benefits large plans; (5) and (6) two

appeals were erroneously factored into CMS’s rating for Plan H3443; and (7) CMS’s reliance on

a private entity to handle appeals from plan coverage denials violates the private nondelegation

doctrine. The Court will reject the first four claims and grant summary judgment to Alignment

as to the two erroneously included appeals. The Court need not reach Alignment’s private non-

delegation claim.

A. Tukey Outlier Rule

The Court rejects Alignment’s claim that the Tukey Outlier Rule is arbitrary and

capricious. As CMS adequately explained, the Tukey Outlier Rule, which underwent notice and

comment in 2023, serves an important purpose: reducing statistical noise by identifying and

removing outlier data points from certain star-rating measures. See A.R. 104;

, 22120 (April 12, 2023). Before adoption of the rule, CMS found that outliers could

“have undue influence on cut points” that CMS uses to divide plans by star-level, potentially

“lead[ing] to a single contract having a major influence on cut point values[.]” A.R. 104. To

address that problem, CMS adopted an accepted method for identifying outliers and removing

them. The natural consequence of removing outliers is some amount of compression in the

remaining dataset and therefore the cut points. But to the extent that such compression results,

3 CMS adequately explained that it is appropriate because it more accurately reflects plan

performance among representative plans.

Alignment urges that the Tukey Outlier Rule should be limited to circumstances where

plans have a sufficiently large denominator of data points to obtain each of the five possible star

ratings. But CMS has justified why it rejected Alignment’s preferred approach. CMS explained

that the situation Alignment posits is relatively rare, as the Tukey Outlier Rule typically does not

result in significant changes to cut points. A.R. 103–04; see 88 Fed. Reg. at 22296. To the

extent it occurs, CMS adequately explained why compression of cut points is nonetheless

appropriate in light of the data reliability concerns the agency was trying to address. So in

CMS’s view, the relatively uncommon circumstance posited by Alignment did not outweigh the

benefits CMS saw in eliminating outliers from the underlying data.

CMS’s adoption of the Tukey Outlier Rule was reasonable and reasonably explained.

The Court will not second-guess it and therefore grants summary judgment to the government

and denies it to Alignment on this issue.

B. Low-Reliability Survey Data

Next, the Court rejects Alignment’s challenge to CMS’s inclusion of low-reliability

survey-response data in composite star-rating measures. As CMS explained, the agency polices

statistical reliability at the measure level by not calculating a star rating for unreliable measures.

See

(a)(3). Doing so ensures that plans get star ratings only for those

measures that reliably reflect plan performance.

Alignment responds that CMS’s decision not to consider reliability at the survey-question

level is arbitrary because unreliable survey response data will necessarily produce an unreliable

composite measure. But as CMS points out, aggregating survey responses that measure or

4 observe a similar subject can result in an overall measure that is more reliable than its constituent

parts. See Gov’t MSJ at 26–27; Health Servs. Advisory Grp., Frequently Asked Questions –

Contracts (Dec. 23, 2024), https://perma.cc/QLJ2-SXZ9. Here, CMS has designed the CAHPS

survey so as to include overlapping questions that can be aggregated to get a more reliable

picture of plan performance. For example, CMS asks multiple questions aimed at assessing the

timeliness of patient appointment scheduling and quality of doctor-patient communication. See

A.R. 399. Finally, CMS does not report question-level response data for composite measures

and uses such data only to inform the ratings it calculates for those measures, so CMS is not

disclosing unreliable information to plan enrollees. See Frequently Asked Questions –

Contracts, supra.

C. Spanish Speakers

The Court also finds that CMS reasonably addressed potential errors in survey

administration that may have caused Spanish-speaking enrollees in Plans H3443 and H3815 to

receive English-language surveys. To begin, Alignment has not shown that the survey was

administered erroneously. CMS survey procedures require that Spanish-language surveys be

“made available to all Spanish-speaking enrollees[.]” A.R. 172. Here, that requirement was

satisfied because, even if Spanish speakers received English-language surveys initially, they

would have been informed in Spanish, either on the survey itself or separately, that Spanish

versions were available upon request. A.R. 172–73; Rough Oral Arg. Tr. 36:18–38:25

(discussing alternative ways of notifying Spanish speakers that Spanish surveys were available);

see A.R. 403–50 (sample notification letters offering Spanish surveys). Alignment points to no

evidence in the record that the requirements were not followed here.

5 Nor did CMS arbitrarily ignore Alignment’s concerns. CMS responded first by ensuring

that DataStat, Alignment’s chosen (and CMS-approved) survey vendor, followed CMS protocols

when matching language-preference data to the samples that CMS drew from Alignment’s

enrollees. A.R. 53. CMS also verified that it sampled Alignment’s plans correctly and that the

samples were consistent with the broader population from which they were drawn. A.R. 48.

And it checked survey response patterns among predicted Spanish speakers and concluded that

they were in line with expected responses. A.R. 61–62. By undertaking this review, CMS took

Alignment’s concerns seriously and responded reasonably. Nothing more was required. See

Petal Gas Storage, LLC v. FERC,

(D.C. Cir. 2007) (An agency “is not

required to choose the best solution, only a reasonable one.”).

D. Oversampling

The Court rejects, for two independent reasons, Alignment’s as-applied challenge to

CMS’s policy of permitting large plans to oversample. First, Alignment’s claim, as presented in

its motion for summary judgment, is an impermissible attempt to amend its complaint by brief.

“It is well settled that a party cannot amend [its] complaint through motions briefing.” Sinha v.

Blinken, No. 20-cv-2814 (DLF),

, at *3 (D.D.C. Sept. 30, 2021) (citing

Durand v. District of Columbia,

(D.D.C. 2014)). Originally,

Alignment’s complaint alleged that permitting oversampling is “irrationally biased against

smaller plans with fewer enrollees” that cannot oversample and that the practice resulted in a

lower rating for Alignment’s H3815 plan. Compl. ¶¶ 153–54. Alignment’s complaint did not

raise this issue as to any of Alignment’s other plans. See

But Plan H3815 is not a small one.

It has over 91,000 enrollees and was therefore eligible for oversampling, which Alignment chose

not to do. Perhaps realizing this contradiction, Alignment now argues that Plan H3815 was

6 disadvantaged not because of its size, but because oversampling irrationally benefits plans that

choose to do it in a way that “does not indicate higher quality.” Alignment MSJ at 40. That

pivot in Alignment’s claim is a significant one and Alignment should have sought leave to

amend its complaint to make it. Having failed to do so, Alignment cannot now try to shoehorn a

new claim into this case.

Second, and in any event, Alignment’s amended claim falls short on the merits.

Oversampling does not guarantee a higher rating, as taking a larger sample of plan enrollees can

confirm a poor rating as much as it can a strong one. To the extent it results in a higher star

rating for certain plans over plans that did not oversample, it does so by increasing the reliability

of the plan’s raw score. Alignment has not shown that it is irrational for HHS to award a higher

star rating to a plan with a more reliable raw score compared to a plan with a less reliable one.

E. Appeals

Finally, the Court agrees with Alignment that CMS failed to adequately explain why it

included adverse determinations in two coverage-denial appeals in Plan H3443’s star rating.

First, CMS’s inclusion of Case 1-13226962526 is irrational. Agency action is arbitrary

and capricious where “there has been a clear error in judgment,” such as when the agency’s

rationale “is so implausible that it could not be ascribed to a difference in view or the product of

agency expertise.” Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co.,

(1983). Here, Alignment erroneously (and belatedly) referred to Maximus an

appeal that Alignment had dismissed rather than denied on the merits. Maximus treated the

appeal as denied, however, and the belated referral therefore negatively affected Alignment’s star

rating. When Alignment objected, CMS declined to reverse the adverse finding on the grounds

that any mistake by Maximus was of Alignment’s own making. See A.R. 1 (“[Y]our contract

7 failed to process this case appropriately[.]”). That rationale, however, makes little sense given

that the government has conceded that “CMS regulations and guidance make clear that plans

should not send dismissed requests to the Independent Review Entity.” Gov’t MSJ at 41. So

Alignment was not required to forward the dismissal to Maximus. If Alignment was not required

to forward the dismissal, then there was no deadline for it to do so. And it is therefore

impossible for Alignment to have forwarded the dismissal to Maximus “late.” That Alignment

nonetheless forwarded the dismissal does not create a deadline that never existed or change the

dismissal into a ruling on the merits.

Second, CMS has abandoned the only reason it gave at the time for including an adverse

determination concerning Case 1-12757246876, an appeal about coordinating care, in

Alignment’s rating. CMS’s sole reason for including this appeal in Alignment’s star rating was

that Alignment failed to seek reopening of this decision before Maximus. A.R. 1. CMS did not

state that it agreed with Maximus’s substantive reasoning, or otherwise express any view on the

merits. See

Before this Court, the government’s motion for summary judgment reiterated

that Alignment failed to seek reopening. Gov’t MSJ at 44. But Alignment then argued in its

reply brief that nothing in CMS’s regulations require a plan provider to seek reopening before

asking CMS to exclude a Maximus decision from its star rating. Alignment Reply at 25. The

government then did not respond to Alignment’s argument in its own reply brief or raise the

issue at oral argument. See Gov’t Reply at 23–24. In doing so, the government has abandoned

the only reason given by CMS for the action it took. See Queen v. Schultz,

(D.D.C. 2015) (holding that because the plaintiff’s reply failed to “make a reference” to an

argument in his opening brief, “the plaintiff has abandoned this argument[,] and the Court need

not consider it.”); Ctr. for Food Safety v. Salazar,

, 152 n.2 (D.D.C. 2012)

8 (“The Court, however, need not address [Plaintiffs’] argument [because] Plaintiffs’ Reply

appears to abandon [it.]”).

“It is a foundational principle of administrative law that judicial review of agency action

is limited to the grounds that the agency invoked when it took the action.” Dep’t of Homeland

Sec. v. Regents of the Univ. of Calif.,

, 20 (2020) (quotation marks omitted). Having

abandoned CMS’s only contemporaneous rationale for including this second appeal in

Alignment’s star rating, the government leaves the Court with “no explanation at all” for CMS’s

actions. CREW v. FEC, No. 22-cv-35 (CRC),

, at *4 (D.D.C. Mar. 17, 2025);

see State Farm,

(“[C]ourts may not accept appellate counsel’s post hoc

rationalizations for agency action.”). The Court therefore agrees with Alignment that CMS’s

inclusion of this second appeal in Plan H3443’s star rating was arbitrary and capricious for

failure to supply an adequate explanation. See State Farm,

(Courts “may not

supply a reasoned basis for the agency’s action that the agency itself has not given.”).1

In light of the above, the Court expresses no view on Alignment’s claim that CMS’s use

of Maximus to handle appeals is unconstitutional under the private nondelegation doctrine. The

only appeals Alignment takes issue with are the two discussed above; it does not claim that any

other appeals were wrongly decided or should not have been adversely considered in its star

ratings. See Compl. ¶ 151 (seeking remand only as to Plan H3443 on private non-delegation

claim). Accordingly, the Court has no need to address this constitutional claim at this time. See

Camreta v. Greene,

(2011) (“[A] longstanding principle of judicial restraint

1 Alignment argues that this second appeal decision is not supported by substantial evidence under

(2)(E). That section, however, applies “only to formal proceedings, not informal adjudications.” Phoenix Herpetological Soc’y, Inc. v. U.S. Fish & Wildlife Serv.,

(D.C. Cir. 2021).

9 requires that courts avoid reaching constitutional questions in advance of the necessity of

deciding them.” (quotation marks omitted)).

IV. Conclusion

For these reasons, the Court will grant in part and deny in part both parties’ motions for

summary judgment. The Court will remand to CMS for recalculation of Plan H3443’s star

ratings without the adverse determinations stemming from the two challenged appeals. A

separate Order follows.

CHRISTOPHER R. COOPER United States District Judge

Date: June 9, 2025

10