HMO Louisiana, Inc. v. Department of Health and Human Services

• Court: Court of Appeals for the D.C. Circuit
• Decided: June 26, 2026
• Docket: 25-5269
• Status: Published

Opinion

United States Court of Appeals FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued February 12, 2026 Decided June 26, 2026

No. 25-5269

HMO LOUISIANA, INC., APPELLANT

v.

DEPARTMENT OF HEALTH AND HUMAN SERVICES, ET AL., APPELLEES

Appeal from the United States District Court for the District of Columbia (No. 1:24-cv-02931)

Paul A. Werner III argued the cause for appellant. With him on the briefs were Imad S. Matini, Hannah Wigger, and Tifenn V. Drouaud.

Kenneth R. Whitley, Attorney, U.S. Department of Health and Human Services, argued the cause for appellees. With him on the brief were Jeanine Ferris Pirro, U.S. Attorney, and Johnny H. Walker III and John Bardo, Assistant U.S. Attorneys. 2 Before: HENDERSON, CHILDS, and PAN, Circuit Judges.

Opinion for the Court filed by Circuit Judge PAN.

Circuit Judge HENDERSON concurs in the judgment.

PAN, Circuit Judge: The Centers for Medicare and Medicaid Services (CMS) administers the Medicare program, which provides government-sponsored healthcare coverage to qualifying beneficiaries. As an alternative to traditional Medicare, CMS allows Medicare beneficiaries to choose coverage provided by private insurers through the Medicare Advantage program. CMS publishes “star ratings” for the private insurance plans that participate in Medicare Advantage, evaluating the plans on a scale of one to five stars. The star ratings are intended to help beneficiaries compare the quality of available insurance plans.

HMO Louisiana, Inc. (HMOLA) is a private insurer that participates in Medicare Advantage. In 2024, HMOLA consolidated two of its Medicare Advantage contracts. When CMS calculated the consolidated contract’s 2025 overall star rating, the agency included certain data that pertained to one of the pre-existing contracts before it was consolidated. HMOLA contends that including that data in the new star-rating calculation violated the governing statute, its implementing regulations, and the agency’s Technical Guidance, and that CMS failed to adequately explain a change in its position in calculating the rating. We disagree on all counts and affirm the district court’s entry of summary judgment in favor of CMS.

I.

Title XVIII of the Social Security Act establishes Medicare, a federally funded and administered program 3 providing health insurance for those over the age of sixty-five and certain disabled persons. 42 U.S.C. § 1395c. The Secretary of the Department of Health and Human Services administers the Medicare program through CMS. The Social Security Act also establishes the Medicare Advantage program — an alternative to traditional, government-managed Medicare. Id. § 1395w-21. 1 Under Medicare Advantage, CMS contracts with private insurers to provide coverage to beneficiaries who otherwise qualify for traditional Medicare. Id. § 1395w-22. Beneficiaries may select private plans available in their geographic areas through Medicare Advantage. Id. § 1395w-21(b).

To assist beneficiaries with choosing insurance plans, Congress has directed CMS to issue annual ratings under a five-star rating system for each plan offered through Medicare Advantage. 2 42 U.S.C. § 1395w-23(o)(4). The star-rating system is designed to provide beneficiaries information that is “a true reflection of [a] plan’s quality and encompasses multiple dimensions of high quality care.” 83 Fed. Reg. 16440, 16520 (Apr. 16, 2018). In addition, star ratings have financial

1 The Medicare Advantage program replaced the Medicare+Choice program in December 2003. Medicare Prescription Drug, Improvement, and Modernization Act of 2003, Pub. L. No. 108–173, § 201, 117 Stat. 2066, 2176. 2 Under the statute, the “contract” refers to the agreement between CMS and a private insurer under which the insurer offers healthcare plans to Medicare beneficiaries. A contract “may cover more than 1 Medicare [Advantage] plan.” 42 U.S.C. § 1395w-27(a); see also 42 C.F.R. § 422.503(a). Per the regulations, CMS calculates the star rating for each contract and assigns that rating to every plan offered under the contract. 42 C.F.R. § 422.162(b)(1)–(2). 4 consequences for insurers: Plans with higher star ratings receive higher “rebate” payments from CMS. 3

The implementing regulations establish a methodology for CMS to calculate a contract’s star rating and provide that CMS may elaborate on that methodology in Technical Notes. 42 C.F.R. §§ 422.162(b), 422.164(a), 422.166. The agency’s Technical Notes contain detailed instructions for calculating ratings for approximately forty individual measures of care, which are components of the overall star rating for each contract. Id. § 422.164(a).

Each measure of care is assigned a “measure score,” which is defined by regulation as a “numeric value of the measure or an assigned ‘missing data’ message.” 42 C.F.R. § 422.162(a). The Technical Notes, in turn, list ten possible missing-data messages, including “Not Applicable” and “No data available.” J.A. 335. Star ratings are calculated using data from two years before the relevant year. Thus, the 2025 star ratings at issue here are based primarily on 2023 measure scores.

As relevant here, a measure of care labeled “C05” evaluates a contract’s “Special Needs Plan (SNP) Care Management.” J.A. 194. SNPs provide benefits and care for

3 Each year, an insurer submits a bid to CMS representing the payment it will accept to cover a beneficiary with an average risk profile. 42 U.S.C. § 1395w-23(a)(1)(B); 42 C.F.R. § 422.254. If that bid falls below a benchmark set by CMS, CMS returns part of the difference to the insurer as a “rebate.” 42 U.S.C. §§ 1395w- 23(a)(1)(E), 1395w-24(b)(1)(C). A plan’s star rating determines the size of that rebate — the higher the rating, the greater the share returned. For example, plans with at least a 4.5-star rating receive seventy percent of the difference between their bid and the benchmark, while those with a 3.5- or 4-star rating receive sixty-five percent of the difference. Id. § 1395w-24(b)(1)(C)(v). 5 beneficiaries who are chronically ill, living in a facility such as a nursing home, or dually eligible for Medicare and Medicaid. According to the Technical Notes, when scoring measure C05, CMS must assign a “Not Applicable” score when the contract at issue does not offer an SNP. J.A. 335. The Technical Notes also direct CMS to assign a “No data available” score when the contract has “an effective termination date on or before the deadline to submit data validation results to CMS.” J.A. 194.

CMS’s regulations also govern the circumstances under which contracts can be consolidated and how CMS calculates a consolidated contract’s star rating during the first two years of its existence. Under the applicable regulations, consolidation occurs “when a [Medicare Advantage] organization that has at least two contracts for health and/or drug services of the same plan type under the same parent organization in a year combines multiple contracts into a single contract for the start of the subsequent contract year.” 42 C.F.R. § 422.162(a). During consolidation, all the beneficiaries enrolled in any consumed contract, i.e., a contract that will no longer exist after consolidation, are transferred to the surviving contract, i.e., a contract that “will still exist under a consolidation.” Id. The star rating for the new consolidated contract is calculated by taking “the enrollment-weighted mean of the measure scores of the surviving and consumed contract(s).” Id. § 422.162(b)(3)(i). 4

4 Specifically, for the first and second years after consolidation, CMS calculates the “enrollment-weighted measure scores using the July enrollment of the measurement period of the consumed and surviving contracts for all measures.” 42 C.F.R. § 422.162(b)(3)(iv)(A)(1), (B)(1). The regulations do provide exceptions to this method for certain measures, but none of those exceptions applies to measure C05. 6 As a Medicare Advantage organization, HMOLA serves approximately 30,000 Medicare beneficiaries in Louisiana through multiple contracts with CMS. In 2024, HMOLA’s parent company consolidated two of its contracts, effective January 1, 2024. In the year before consolidation (2023), the consumed contract offered an SNP, which was rated under measure C05. The surviving contract did not offer an SNP and therefore had no data available for measure C05. After consolidation, the new consolidated contract offered an SNP for 2025.

In September 2024, CMS initially gave HMOLA’s consolidated contract a 3.5 overall star rating for 2025. In that calculation, CMS did not include a score for measure C05. CMS categorized the consumed contract as having been terminated and therefore interpreted the Technical Notes to require exclusion of its C05 data. But HMOLA believed that CMS erred by excluding the consumed contract’s C05 star rating, and that including it would increase the consolidated contract’s overall rating from 3.5 to 4 stars. Thus, HMOLA notified CMS of the perceived error and requested that CMS include the consumed contract’s C05 data in the overall star- rating calculation. HMOLA also noted that it had been “unable to submit relevant measurement information data” for measure C05 through CMS’s data-submission portal. J.A. 364.

CMS initially denied HMOLA’s request to include the consumed contract’s C05 score. HMOLA then made additional arguments, explaining that excluding the consumed contract’s C05 data conflicted with applicable regulations and Technical Notes.

HMOLA’s lobbying was successful. CMS ultimately decided to accept the consumed contract’s C05 data and to include it in the consolidated contract’s star-rating calculation, 7 noting that it did so “[a]t [HMOLA]’s request and after further consideration.” J.A. 375. The agency reprogrammed its data- submission portal to conform to this change and to allow HMOLA to submit the consumed contract’s C05 data.

When CMS recalculated the consolidated contract’s overall star rating, it included measure C05. For that measure, it used “the enrollment-weighted mean of the measure scores of the surviving and consumed contract(s).” 42 C.F.R. § 422.162(b)(3)(i). The two scores it averaged were: (1) a “Not Applicable” C05 score for HMOLA’s surviving contract, which did not offer an SNP in 2023, and (2) the C05 score for the consumed contract for 2023, which was seventy percent, or 3 stars. CMS determined that the enrollment-weighted mean of those values was seventy percent, or 3 stars. When the C05 score was considered in calculating the overall rating for the consolidated contract, the overall rating was unchanged — it was still 3.5 stars.

Disappointed with those results, HMOLA sued CMS under the Administrative Procedure Act (APA), challenging the recalculation method that it had urged CMS to adopt. Due to other changes that CMS had implemented in the recalculation, HMOLA believed that excluding the C05 data would now result in a higher overall star rating for the consolidated contract, even though the original calculation that had excluded measure C05 had yielded the same rating of 3.5 stars. 5 Before the district court, HMOLA claimed that the

5 HMOLA initially asked CMS to include data from both measure C05 and another measure, D11, from the consumed contract. CMS granted both requests during its recalculation. While the consumed contract’s data improved the consolidated contract’s D11 score, the C05 data proved lower than HMOLA had anticipated. Consequently, the overall rating remained 3.5 stars — the same rating the original calculation produced when CMS excluded both 8 agency’s methodology for recalculating the consolidated contract’s star rating was arbitrary and capricious, in violation of the APA. HMOLA took a position that was the opposite of what it had previously argued, asserting that the inclusion of the consumed contract’s C05 rating in calculating the consolidated contract’s 2025 star rating contravened the regulations, the governing statute, and the Technical Notes.

After the parties filed cross-motions for summary judgment, the district court ruled in favor of CMS. The district court concluded that the plain language of the governing statute, regulations, and Technical Notes supported CMS’s inclusion of the consumed and surviving contracts’ 2023 C05 scores in the consolidated contract’s star-rating calculation. It also concluded that CMS’s decision to recalculate the overall rating to include the consumed contract’s C05 score did not constitute a change in policy that required further explanation. Alternatively, it held that even if an explanation were required, the circumstances made clear that CMS had adopted HMOLA’s reasoning for including the consumed contract’s C05 data.

HMOLA appealed. We have jurisdiction under 28 U.S.C. § 1291.

II.

The APA requires a reviewing court to set aside agency action that is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A). Whether an agency action is arbitrary and capricious is a legal question “generally made on the administrative record and

datasets. It appears that HMOLA can achieve a higher overall rating only if CMS retains the favorable D11 data while omitting the unfavorable C05 data. 9 resolved on summary judgment.” Ctr. for Biological Diversity v. Zeldin, 171 F.4th 356, 376 (D.C. Cir. 2026). We review such a determination de novo. Id.

An agency action that violates the agency’s own regulations or a statute may be set aside as arbitrary and capricious. Ctr. for Biological Diversity, 171 F.4th at 376; see also Nat’l Env’t Dev. Ass’n’s Clean Air Project v. EPA, 752 F.3d 999, 1009 (D.C. Cir. 2014). An agency action is likewise arbitrary and capricious when the agency departs from prior policy without a reasoned explanation. Encino Motorcars, LLC v. Navarro, 579 U.S. 211, 222 (2016).

III.

HMOLA mounts three challenges to CMS’s recalculation of the consolidated contract’s 2025 overall star rating. First, HMOLA contends that the agency’s calculation methodology conflicts with CMS’s regulations and Technical Notes. Second, it argues that CMS violated its statutory mandate to provide beneficiaries with useful information about a Medicare Advantage plan’s quality. Third, it claims that CMS failed to adequately explain its change in position when it recalculated the 2025 star rating. We find those arguments unpersuasive.

A.

CMS properly applied the plain text of the applicable regulation when it calculated the star rating of the consolidated contract at issue. The regulation provides that for the first two years following consolidation, a consolidated contract’s star rating is “the enrollment-weighted mean of the measure scores of the surviving and consumed contract(s).” 42 C.F.R. § 422.162(b)(3)(i); see also id. § 422.162(b)(3)(iv)(A)(1), (B)(1). Consistent with that regulation, “[i]f neither the consumed nor the surviving contract offered a[n] [SNP] in the 10 measurement year, CMS does not include measure C05 in a contract’s overall score.” Gov’t Br. 24; see also id. at 23–24 (citing 42 C.F.R. § 422.162(b)(3)(i), (iv)). But if either of the pre-existing contracts received a C05 score, and the consolidated contract continues to offer an SNP, CMS uses the enrollment-weighted average of the previous scores to calculate the consolidated contract’s rating. 42 C.F.R. § 422.162(b)(1).

Here, CMS followed that methodology when it recalculated the consolidated contract’s 2025 star rating at HMOLA’s request. The agency included as inputs the consumed contract’s 3-star rating for measure C05 in 2023, and the surviving contract’s “Not Applicable” rating for that year. CMS calculated the enrollment-weighted mean of those values to be 3 stars. And that measure score for C05 was considered in calculating the consolidated contract’s overall rating of 3.5 stars.

Although HMOLA previously endorsed the above- described plain-text application of the relevant provisions, it now takes a contrary position. Its new theory is that a consumed contract “no longer exist[s]” upon consolidation, and therefore is effectively “terminated.” HMOLA Br. 27 (quoting 42 C.F.R. § 422.162(a)). HMOLA claims that the consumed contract’s C05 score should thus be excluded as “terminated,” based on a provision in CMS’s Technical Notes, which states that a terminated contract’s data should be “excluded and listed as ‘No data available’” for the C05 measure. J.A. 194 (“Contracts . . . with an effective termination date on or before the deadline to submit data validation results to CMS (June 15, 2024) are excluded and listed as ‘No data available.’”). Excluding the consumed contract’s C05 data under this theory would result in no data at all for measure C05, and that would require measure C05 to be 11 omitted from the consolidated contract’s star-rating calculation.

We discern at least two problems with HMOLA’s analysis. First, contrary to HMOLA’s assertion, a “consolidation” does not result in the “termination” of a “consumed” contract under the regulations and Technical Notes. The applicable regulation provides that “consolidation” occurs “when [a Medicare Advantage] organization . . . combines multiple contracts into a single contract for the start of the subsequent contract year.” 42 C.F.R. § 422.162(a). Furthermore, the regulation defines a “consumed contract” as “a contract that will no longer exist after a contract year’s end as a result of a consolidation.” Id. Although a consumed contract ultimately is absorbed into the consolidated contract, it is not considered “terminated.”

“Termination” under the regulations addresses the ability of CMS and a Medicare Advantage organization to end their contractual relationship. A Medicare Advantage contract may be “terminated” only (1) “by mutual consent” of the parties, 42 C.F.R. § 422.508; (2) by CMS if the agency “determines that the [Medicare Advantage] organization” has neglected its duties under the contract or has committed fraud, id. § 422.510; or (3) by the Medicare Advantage organization “if CMS fails to substantially carry out the terms of the contract,” id. § 422.512. 6 The regulations do not include “consolidation” within their discussion of “termination.”

The Technical Notes reinforce the distinction between consolidated and terminated contracts. While the Technical Notes establish that CMS’s rating calculation should exclude

6 HMOLA insists that the list of circumstances triggering a “terminated” contract is non-exhaustive. Its reasoning, however, is circular: HMOLA claims the list must be non-exhaustive because it does not include “consolidation” as a method for termination. 12 the C05 data of a “terminated” contract, the Notes specifically mandate different treatment for “consolidated” contracts. Consistent with the regulation, 42 C.F.R. § 422.162(b)(3)(i), the Technical Notes provide that “the measure values for the surviving contract of a consolidation are calculated as the enrollment-weighted mean of all contracts in the consolidation.” J.A. 271. 7

HMOLA relies on a dictionary definition of uncertain provenance to assert that “terminated” simply means “having come or been brought to an end or conclusion.” HMOLA Br. 27 n.7. 8 But even if a dictionary definition illuminates the ordinary meaning of a term, such a definition cannot override the regulatory text or structure. Cf. Pac. Gas & Elec. Co. v. FERC, 113 F.4th 943, 948 (D.C. Cir. 2024) (“We assume that ‘statutory terms bear their ordinary meaning’ unless evidence suggests otherwise.” (quoting Niz-Chavez v. Garland, 593 U.S. 155, 163 (2021))). Where, as here, the regulations distinguish between “consolidation” and “termination,” a dictionary definition cannot supplant the regulatory framework.

7 The parties dispute whether the Technical Notes carry the same binding authority as that of the governing regulations, and whether the Technical Notes’ provision regarding exclusion of data for terminated contracts supersedes the general rule requiring that a consolidated contract’s rating be determined by calculating the weighted average of the consumed and surviving contracts’ scores. We need not reach that question because the regulations and the Technical Notes are not in conflict. Rather, they point in the same direction: Both require CMS to include the consumed contract’s C05 data in its calculation of HMOLA’s consolidated-contract rating. 8 HMOLA cites the “Oxford Mod. Dictionary (2025)” for this definition. HMOLA Br. 27 n.7. Although dictionaries with similar titles exist, we have been unable to locate an edition matching that citation or to confirm the quoted definition in Oxford’s available sources. 13 In sum, we reject HMOLA’s argument that a “consumed” contract should be treated as a “terminated” contract for purposes of calculating the consolidated contract’s star rating. CMS’s inclusion of the C05 data in its rating calculation was consistent with the applicable regulations and Technical Notes. 9

B.

HMOLA next argues that CMS’s calculation fails to assist beneficiaries in making informed selections of Medicare Advantage plans because the agency’s 2025 rating for the consolidated contract is somehow inaccurate. According to HMOLA, the disputed calculation does not “provide information to the beneficiary that is a true reflection of the plan’s quality,” as required by the statute, because it does not

9 We note that HMOLA has not challenged the math that CMS used to calculate the weighted average of the two relevant data points, i.e., the 2023 C05 rating of 3 stars for the consumed contract and “Not Applicable” for the surviving contract. CMS determined that the weighted average of those values was 3 stars. After oral argument, we ordered the parties to address in supplemental briefing whether CMS correctly applied the statute, 42 U.S.C. § 1395w- 23(o)(4)(D)(i), and implementing regulation, 42 C.F.R. § 422.162(b)(3)(i), when it determined that the enrollment-weighted mean of the C05 score for the consumed contract (3 stars) and the surviving contract (Not Applicable) was 3 stars. HMOLA, however, failed to answer the question that we posed. Instead, in its supplemental briefing, HMOLA rehashed its argument that its consumed contract should have been deemed “terminated,” and its corresponding C05 data excluded. Because HMOLA’s opening and reply briefs challenge only the inclusion of the C05 data in the star- rating calculation, and because HMOLA declined our invitation to argue in a supplemental brief that the mathematical calculation of the weighted average was incorrect, it has doubly forfeited any challenge to CMS’s math. 14 properly account for the fact that one of the pre-existing contracts did not offer an SNP. 83 Fed. Reg. at 16520. We disagree.

As discussed, the consolidated contract comprises a “consumed” contract and a “surviving” contract. Here, the “consumed” contract offered an SNP and was rated on measure C05, but the “surviving” contract did not offer an SNP. Because the new consolidated contract does offer an SNP, the quality of the consumed contract’s SNP services is relevant to the consolidated contract’s star-rating calculation. That the surviving contract did not offer an SNP was properly considered: As required by the applicable regulation, CMS averaged the consumed and surviving contracts’ scores in calculating the consolidated contract’s rating. 42 C.F.R. § 422.162(b)(3)(iv)(A)(1), (B)(1). CMS thus provided beneficiaries with accurate information regarding the quality of HMOLA’s consolidated contract. 10

C.

HMOLA’s final argument is that CMS changed its policy by including the C05 data in its recalculation of the 2025 rating, and that CMS failed to adequately explain its reason for adopting a new approach. When CMS announced that it would accept the C05 data, it stated only that it was doing so “at [HMOLA]’s request and after further consideration.” J.A. 375. Although CMS did not offer a detailed explanation of why it recalculated the 2025 rating for the consolidated contract, the

10 HMOLA suggests that CMS waived any argument that using the consumed contract’s C05 data provides a more accurate reflection of the consolidated contract’s quality. Not so. CMS raised this argument below in its summary-judgment briefing. See J.A. 77. 15 recalculation did not reflect a change in policy that required explanation.

The change-in-position doctrine requires that an agency “provide a reasoned explanation” when it changes an existing policy. FDA v. Wages & White Lion Invs., LLC, 604 U.S. 542, 568 (2025) (cleaned up). An agency changes an existing policy when, for example, it “acts inconsistently with an earlier position,” “disavows prior inconsistent agency action as no longer good law,” or abandons a “decades-old practice.” Id. at 569–70 (cleaned up). The doctrine serves to ensure that, before reversing course, an agency remains “cognizant that [its] longstanding policies may have engendered serious reliance interests.” Id. at 570 (cleaned up). 11 We have accordingly required an agency to provide a reasoned explanation when it departs from a settled practice or publicly articulated policy. See Am. Wild Horse Pres. Campaign v. Perdue, 873 F.3d 914, 927 (D.C. Cir. 2017) (holding that an agency cannot depart from its “established pattern of agency conduct and formalized positions” without a reasoned explanation); see also Commc’ns & Control, Inc. v. FCC, 374 F.3d 1329, 1335 n.8 (D.C. Cir. 2004) (noting that an agency’s unexplained departure from its longstanding practice is arbitrary and capricious).

CMS’s recalculation of the consolidated contract’s star rating does not amount to a change in “policy” that requires “reasoned explanation.” Although CMS did change its methodology in response to HMOLA’s objection, the record does not reflect that the original calculation implemented a pre-

11 The Supreme Court has “traditionally applied the change-in- position doctrine when an agency shifts from a position expressed in a more formal setting,” and it has “assume[d], without deciding, that the change-in-position doctrine applies to an agency’s divergence from a position articulated in nonbinding guidance documents.” Wages & White Lion, 604 U.S. at 569 n.5. 16 existing “policy.” To the contrary, CMS explained that the agency had never before considered whether to include a consumed contract’s C05 data in a post-consolidation rating. See J.A. 138 (CMS’s counsel explaining to the district court that “[t]here are not that many contract consolidations that involve special needs plans, and, of course, it’s even less likely that there’s a consumed contract that offers special needs plans, and a surviving contract [that] does not”). Thus, a more detailed explanation of the agency’s action in this instance was not required.

Notably, the revised calculation did not upset any “serious reliance interests.” Wages & White Lion, 604 U.S. at 570. Instead, the recalculation occurred during a preliminary review process, in which CMS allowed HMOLA to provide feedback before CMS finalized the consolidated contract’s star rating. See 42 C.F.R. § 422.166(h)(2) (establishing that Medicare Advantage providers “can preview their preliminary Star Ratings data . . . prior to display on the [public] Medicare Plan Finder”); see also J.A. 364 (HMOLA’s email to CMS during the review process: “Thank you for allowing [us] the opportunity to provide feedback on the Star Rating calculation prior to finalization.”). During that preliminary process, it was HMOLA that persuaded CMS to adjust its methodology to include the consumed contract’s C05 data. Having received exactly what it requested, HMOLA can hardly claim that it was surprised by the outcome or that the agency upset its reliance on CMS’s initial calculation. Under these circumstances, CMS’s recalculation plainly is not a change in policy that required reasoned explanation, and we therefore need not decide whether CMS adequately explained its decision to include the C05 data. 12

12 HMOLA also points to the fact that CMS had to reprogram its data-submission portal to accept the consumed contract’s C05 data, 17 * * *

For the foregoing reasons, we affirm the district court’s entry of summary judgment in favor of CMS.

So ordered.

suggesting that the portal’s original configuration reflected a past practice of exclusion. But HMOLA cites no authority to support the novel proposition that CMS announced a binding policy solely by configuring its data-submission platform in a certain way.