Appeal from the United States District Court for the District of Columbia (No. 1:25-cv-00074)

Michael B. Kimberly argued the cause for appellant. With him on the briefs was Edward A. Day .

Jack Starcher , Attorney, U.S. Department of Justice, argued the cause for appellees. With him on the brief were Brett A. Shumate , Assistant Attorney General, and Courtney L. Dixon , Attorney.

Before: SRINIVASAN, Chief Judge , KATSAS, Circuit Judge , and ROGERS, Senior Circuit Judge .

Opinion for the Court by Senior Circuit Judge ROGERS.

Page 2ROGERS, Senior Circuit Judge : Alignment Healthcare is a health insurance company that offers Medicare Advantage plans, which are funded and regulated by the Centers for Medicare & Medicaid Services (“CMS”), in the Department of Health and Human Services, 42 U.S.C. § 1395kk(a) (2024). Medicare Advantage plans are subject to a rating system in which they are evaluated based on several metrics, including an annual survey of enrollees. Id. § 1395w-23(o)(4)(A). In September 2024, Alignment contacted CMS claiming that the ratings for two of its plans were inaccurate because some Spanish-speaking customers received surveys in English, contrary to Alignment's request of their preference. After examining the response data and consulting the survey vendor, CMS denied Alignment's request to disregard the survey results. Alignment challenged that decision pursuant to the Administrative Procedure Act (“APA”), and the district court granted summary judgment in relevant part for CMS.

On appeal, Alignment contends the refusal to discard survey data infected by a clear survey administration error was arbitrary and capricious because the refusal to enforce the CMS protocols treated Alignment differently from other Medicare Advantage plan sponsors whose ratings scores relied on properly administered surveys. Appellant's Br. at 25. This court affirms. Even if CMS protocols made Alignment's request to distribute Spanish-language questionnaires binding on the vendor, Alignment fails to establish that its request was disregarded. Similarly, to the extent Alignment challenges the adequacy of CMS’s action or explanation, that challenge fails because CMS examined Alignment’s concerns and explained why there was no indication of survey administration error.

Page 3I.

The Medicare Act, 42 U.S.C. § 1395 et seq. , requires health insurance for individuals who are over the age of 65 or who have certain disabilities. Under Medicare Parts A and B, CMS acts as the insurer, paying healthcare providers directly for beneficiaries’ medical services. UnitedHealthcare Ins. Co. v. Becerra , 16 F.4th 867, 872 (D.C. Cir. 2021); see 42 U.S.C. § 1395c-1395i-6 (Part A); id. § 1395j-1395w-6 (Part B). Medicare Part C allows beneficiaries to receive health insurance through private Medicare Advantage plans, see id. § 1395w-22(a), administered by private health insurers, which contract with CMS to provide coverage in particular areas of the United States. Id. § 1395w-21(b). The insurance providers receive monthly payments from CMS to provide coverage for each beneficiary. Id. § 1395w-23(a)(1).

Medicare Advantage contracts, which include one or more Medicare Advantage plans, id. § 1395w-27(a), are subject to a five-star rating system. Id. § 1395w-23(o)(4)(A). The ratings assist beneficiaries in choosing plans and are also used in calculating the monthly funding received by the insurance providers. Id. § 1395w-24(b)(1)(C)(v). To prepare the ratings, CMS evaluates data from a range of sources, including the Consumer Assessment of Healthcare Providers & Systems (hereinafter, 'the survey'). See 42 C.F.R. § 422.166. The survey is based on information from a sample of enrollees in each Medicare Advantage plan about their experiences. Medicare Advantage contracts with at least 600 enrollees are required to use CMS-approved vendors to administer the survey.

CMS requirements for survey administration are set forth in the Quality Assurance Protocols & Technical Specifications, Page 4V14.1 (Nov. 2023) (hereinafter, “Protocols”). The Protocols include several requirements regarding survey administration in Spanish, including that 'Spanish language questionnaires must be made available to all Spanish-speaking enrollees (in web, mail, and telephone administration).” Id. at 50. When administering the survey by mail, vendors must include a “prenotification letter” advising of the survey, and this letter must be printed in English on one side and Spanish on the other. Id. at 33–34. When administering the survey online, vendors are “required to administer the web survey in English and Spanish,” id. at 27, with recipients able to choose between English or Spanish upon initiating the survey, id. at 30. The survey vendor, in turn, 'may' use a variety of additional strategies “at the request” of the insurance provider in distributing the survey in non-English languages. Id. at 50. Such strategies include sending 'web survey invitations in Spanish only to enrollees known to prefer Spanish' and 'a Spanish language questionnaire only in all mailings of the survey to enrollees known to prefer Spanish' or mailing questionnaires in both Spanish and English. Id. at 50-51.

To conduct the survey, CMS compiles a random sample of enrollees from a Medicare Advantage contract. Id. at 17. Excluded from the sample are those who fail to meet certain criteria, such as individuals who have not been continuously enrolled in the plan for six months. Id. CMS provides the sample file of enrollees to the vendor, including 'mailing addresses of enrollees for whom addresses are available' in CMS's Integrated Data Repository. Id. at 17-18. Before mailing survey materials, “vendors must use commercial tools, such as the [National Change of Address] database, to update addresses provided by CMS for sampled enrollees and to standardize addresses to conform to U.S. Postal Service formats.' Id. at 40. Throughout the administration of the survey, vendors must “maintain the confidentiality of enrollees Page 5and may not provide contracts/plans with the names of enrollees selected for the survey or any other enrollee information that could be used to identify an individual sampled enrollee.' Id. at 50.

After the vendors administer the survey, CMS provides two plan preview periods. 42 C.F.R. § 422.166(h)(2). In the first period, insurance providers can “review the methodology and their posted numeric data for each measure.” 83 Fed. Reg. 16,440, 16,588 (Apr. 16, 2018). In the second period, insurance providers can review “any revisions made as a result of the first plan preview” and their “preliminary Star Ratings.” Id. Insurance providers can send comments to CMS by email following this review. CMS Mem. at 3 (Oct. 26, 2023).

On September 13, 2024, Alignment contacted CMS to dispute the preliminary ratings in the second preview period for two of its contracts (Nos. H3443 and H3815). Email at 27–28 (Sept. 13, 2024). Because the 'Spanish survey responses dropped significantly” for these two contracts between 2023 and 2024, Alignment claimed the results reflected either (1) a sampling error that caused a lower percentage of Spanishspeaking members to be selected to take the survey, or (2) a survey administration error in which some “Spanish-speaking members may not have received the Spanish version of the survey, despite this being their preferred language.' Id. Either error, Alignment claimed, would have 'had a significant negative impact on our . . . scores” because Spanish-speaking enrollees reported 'at least 10% higher satisfaction rates compared to English-speaking members' in Alignment's internal surveys. Id. Alignment requested that 'CMS conduct a thorough examination of our samples or suppress the [survey] results for these contracts.' Id. at 28. Alignment also requested Page 6data regarding the prior year’s survey. Email at 21–22 (Sept. 16, 2024).

In its initial response, CMS confirmed that the surveys 'drew a random sample' of enrollees for each contract and provided data comparing respondents who primarily spoke Spanish at home with those who completed the Spanish language survey. CMS Email at 22–23 (Sept. 16, 2024). The data showed that about a quarter of respondents for each contract reported primarily speaking Spanish at home and about a fifth of respondents completed the survey in Spanish. Id. CMS stated that the results were 'consistent with choices made by members for whom both Spanish and English surveys were available' and that the Protocols required 'that at a minimum, Spanish surveys be made available upon request from members in response to bilingual prenotification materials.” Id. at 23. In further response, CMS shared the figures for 2023–2024 showing (1) the percentage of respondents 'who answered the survey in Spanish in each year” and (2) those “with a high (>30%) predicted probability of Spanish preference,' and stated 'both the percentage of respondents who completed the Spanish language survey and the percentage of respondents likely to prefer Spanish dropped” between the two years for both contracts. CMS Email at 21 (Sept. 17, 2024).

Alignment responded that the data indicated a problem with the survey results because it showed a “significant decline in both predicted Spanish preference rates and Spanish survey completions among survey responses” despite “a slight yearover-year increase in the percentage of Spanish-speaking [survey]-eligible members in the files . . . submitted to our vendor.” Email at 19 (Sept. 17, 2024). It requested that CMS provide data covering 'the percentage of high predicted probability of Spanish preference within the entire [survey]- Page 7eligible population compared to the selected 800-sample,' which would help to 'determine if there is any bias in the sample itself,' or 'point to an issue with the matching or fulfillment process” if bias was not found. Id. Alternatively, Alignment stated that it was 'open to a swift and fair resolution”: CMS could either “suppress the [survey] results for these contracts and calculate our overall performance without including these scores,' or 'mark[] . . . two of the [survey] measures” as having “Very Low Reliability.” Id. at 20.

CMS stated in response the requested data showed that '[i]n each year in each contract the proportion of sampled enrollees with high predicted Spanish probability was similar to the proportion among the larger set of [survey] eligibles from which the sample was drawn.” CMS Email at 16 (Sept. 19, 2024). Noting it was “common for patient surveys” to feature lower response rates from Spanish speakers than from survey recipients as a whole, and that it was “typical” for a portion of those preferring Spanish to take the survey in English, id. , CMS concluded the data was “consistent with a random sample and typical patterns of survey nonresponse,” id . at 17.

Alignment repeated its concerns, including that a “marked reduction in Spanish-language survey completions” suggested that “Spanish-speaking members may have received surveys in English” or “errors in sample selection.” Email at 14–15 (Sept. 20, 2024). CMS responded, stating that it had 'looked extensively into these concerns” and had not found evidence of errors in how the sample was compiled or how the survey data was analyzed. CMS Email at 12 (Sept. 23, 2024). “[A]s such,” CMS stated, it had “no authority to remove any of the [survey] data from the contract's overall Star Rating.' Id. Further, “[w]hether or not Spanish-speaking members received surveys in English despite their preference is outside of CMSPage 8 control. . . . CMS does not get involved in how survey vendors implement language preference data. This is determined by the plan and their vendor working together.' Id. at 13.

Alignment continued to express concern about the Spanish-language responses, Email at 9 (Sept. 23, 2024), and CMS reiterated it had verified that the sample was representative of Spanish speakers and that the responses “were as expected and consistent with the correct administration of the survey.” CMS Email at 8 (Sept. 25, 2024). CMS pointed out that “[r]espondents who primarily speak Spanish at home are often also able to complete the survey in English and many choose to do so when both Spanish and English survey options are made available, as was the case here.' Id. Also, the proportion of respondents speaking primarily Spanish at home filled out surveys in English at lower rates than for Medicare Advantage plans overall, which indicated 'unusually good access to Spanish-language surveys' for the two contracts. Id. In addition, CMS stated the survey vendor had 'attested' to following the Protocols and using 'the language preference data shared by the plan.' Id. The vendor stated that it “matched [the language preference data] as well as possible to the CMS sample file' using 'components such as name and detailed components of the address to be sure the match is correct.” DataStat Email at 1 (Sept. 18, 2024).

When CMS again denied a request for reconsideration, Alignment asked for permission 'to survey our Spanishspeaking [survey]-eligible members directly” about “whether they received the [ ] survey this year and in what language.” Email at 4 (Sept. 26, 2024). CMS refused, repeating that its “analyses have established that the [survey] sample represented Spanish-preferring members, used the language preference information you provided, and resulted in Spanish-preferring members choosing to respond in Spanish and rates that were Page 9high and higher than average.' CMS Email at 3 (Sept. 30, 2024). 'No further validation is needed.' Id. CMS explained that a separate survey “would not be permitted” in view of the CMS Protocols on respondent confidentiality, and “would not produce reliable information” by “asking about a survey that . . . may or may not have [been] received 6 to 8 months ago.” Id.

On January 10, 2025, Alignment filed suit, challenging CMS’s order denying relief on multiple grounds pursuant to the APA. On cross motions for summary judgment, the district court, as relevant, granted summary judgment for the Department. Alignment appeals.

This court reviews de novo the district court's grant of summary judgment Dist. Hosp. Partners, L.P. v. Burwell , 786 F.3d 46, 54 (D.C. Cir. 2015). The court will “affirm summary judgment for the agency unless it violated the Administrative Procedure Act by taking action that is arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” Id. (citation and internal quotation marks omitted). This standard examines whether the agency’s “decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment.” Marsh v. Oregon Nat. Res. Council , 490 U.S. 360, 378 (1989) (citation omitted). The 'deferential standard of review' is particularly deferential where agency decisions 'involve complex judgments about sampling methodology and data analysis that are within the agency’s technical expertise,” so long as the agency applies its “expertise in a reasoned manner.” Dist. Hosp. Partners, 786 F.3d at 60 (citations omitted). 'The party challenging an agency’s action as arbitrary and capricious bears the burden of Page 10proof.” City of Olmsted Falls v. FAA , 292 F.3d 261, 271 (D.C. Cir. 2002) (citation and alteration omitted).

Alignment's primary contention is that CMS departed from the Protocols by refusing to discard the survey results for two Medicare Advantage plan contracts, thereby violating the requirement that “agencies must ‘follow their own rules, even gratuitous procedural rules that limit otherwise discretionary actions.’” Appellant’s Br. at 28 (quoting Steenholdt v. FAA , 314 F.3d 633, 639 (D.C. Cir. 2003)). Alignment relies on the requirements in the Protocols that “Spanish language questionnaires must be made available to all Spanish-speaking enrollees (in web, mail, and telephone administration),' and that vendors may take several different approaches ' at the request of the contract ” to administer the survey in Spanish, such as by mailing “a Spanish language questionnaire only” to recipients 'known to prefer to Spanish.' Id. at 13, 30 (emphasis in original) (quoting Protocols at 50-51). These provisions, Alignment maintains, required the vendor to send Spanish-only materials to enrollees identified as preferring Spanish, and the vendor violated this requirement by sending English surveys to 'a material number' of enrollees who preferred Spanish, approximately 15% of Spanish-speaking enrollees in one plan, and 10% of such enrollees in another. Id. at 31.

Even assuming that the Protocols made Alignment's Spanish-materials request binding on the vendor, and that CMS would be required to discard a survey administered in violation of the Protocols, Alignment fails to show that CMS acted arbitrarily or capriciously contrary to the request or the Protocols. The vendor informed CMS that it had “received a preferred language file' from Alignment 'and matched it as Page 11well as possible to the CMS sample file received from each plan,” using “components such as name and detailed components of the address to be sure the match is correct.” DataStat Email at 1. CMS could reasonably conclude that the vendor did its best to send Spanish questionnaires to those whose contact information from the two data files was the same, while refraining to recognize an individual in the sample file as the same as the one in the language preference data if there were mismatches in contact information, under a “very conservative approach' to data-matching. Id . That conclusion is consistent with the survey data showing that Spanishspeaking enrollees responded in Spanish at 'rates that were high and higher than average.' CMS Email at 3 (Sep. 30, 2024). Therefore, Alignment fails to point to evidence in the administrative record indicating that the vendor refused to follow its request. To the extent Alignment is objecting that the vendor may have defaulted to sending English questionnaires when it could not match Alignment’s data with CMS’s sample file, Alignment identifies no provision of the Protocols that would prohibit such a data-matching approach.

Neither has Alignment shown that it was unreasonable for the vendor to employ “a conservative approach” to resolving mismatches or for CMS to accept the survey results. Alignment has not indicated that it requested the vendor to “[i]nclude a Spanish language questionnaire in all mailings of the English language questionnaire,' which the Protocols identify as an option, at 50. Even if the vendor did not use a Spanish-language default for mismatches, the CMS Protocols impose several requirements on vendors that would make a Spanish-language questionnaire “available” to Spanishspeaking enrollees. In administering the survey by mail, vendors must mail a pre-notification letter that is “printed with English on one side and Spanish on the other side” and that includes the “vendor’s toll-free telephone number for sampled Page 12enrollees to call to request a Spanish language survey,” which 'must be mailed within two days of the telephone request.' Id. at 51. Vendors are “required to administer the web survey in English and Spanish.” Id. at 27. Such requirements enable individuals to have access to a Spanish-language survey through multiple methods. Alignment’s objection that there is 'no evidence' in the administrative record that the vendor followed these requirements, Appellant’s Br. at 42–43, overlooks that Alignment — not CMS — bears the burden of showing that CMS's action was arbitrary and capricious, Olmsted Falls , 292 F.3d at 271.

Without a foothold in the Protocols, Alignment turns to three additional contentions, but none is persuasive. Neither is its challenge to CMS’s investigation or explanation.

“Typically, a court cannot affirm an agency decision on a ground other than those relied upon by the agency.” Wilson v. Fed. Mine Safety & Health Rev. Comm’n , 863 F.3d 876, 883 (D.C. Cir. 2017) (citing SEC v. Chenery Corp. , 318 U.S. 80, 88 (1943)). Alignment contends, first, that this court cannot affirm on the basis that it failed to show a violation of the Protocols because this was “not the rationale that CMS gave during the plan preview periods.” Appellant’s Br. at 39; see Reply Br. at 9-13. Yet a party forfeits its Chenery contention if it fails to raise it in the district court, absent 'exceptional circumstances.' Byers v. Comm'r of Internal Revenue , 740 F.3d 668, 681 (D.C. Cir. 2014) (citation omitted). Forfeiture applies here.

According to Alignment, CMS 'responded to [its] concerns by insisting that no language-mismatch error had occurred and that . . . survey issues were not its problem.'Page 13 Appellant’s Br. at 39. Under this view, CMS could not argue that its actions were consistent with the Protocols’ requirements because it failed to invoke this reasoning in its email correspondence with Alignment. But Alignment first raised its contention that CMS’s inaction violated the Protocols in the district court. In correspondence with CMS, Alignment had stated that it had “identified serious discrepancies in the [ ] survey results that threaten the accuracy and validity of our Star Ratings.” Email at 14 (September 20, 2024). In the district court, Alignment went beyond argument regarding the 'accuracy and validity' of the survey results, claiming that “CMS’s approved survey vendors are required by CMS's own guidelines to make Spanish-language questionnaires available” and that Alignment’s vendor failed to administer the survey “in a manner consistent with” the Protocols. Mem. of Points and Authorities in Support of Plaintiff’s Mot. for Sum. Judgment at 45. In opposing Alignment's motion, CMS argued: (1) the Protocols 'require that all enrollees have the opportunity to complete the survey in Spanish, regardless of whether they were sent a Spanish-language paper survey initially,” Opposition at 40; (2) the initial receipt of a Spanish-language survey concerned 'Alignment's coordination with its own survey vendor,' which did not 'fall within CMS's direct oversight or control,' id. at 39; and (3) Alignment's vendor had told CMS that it had “matched the client data to enrollee names and addresses,” while taking “a conservative approach to this process and refrain[ing] from matching records when there are any conflicts between names or addresses in the plan’s data and CMS’s sample data,” id. at 36. In response, Alignment did not raise Chenery or argue that the district court could not affirm CMS based on these arguments. See Opposition & Reply at 15–18.

No exceptional circumstances excuse Alignment’s failure to raise a Chenery objection in the district court. See Byers ,Page 14 740 F.3d at 681. Alignment had ample time during the course of its email exchange with CMS to present arguments about what the Protocols required, particularly after CMS's initial response was that the Protocols “require that at a minimum, Spanish surveys be made available upon request from members in response to bilingual prenotification materials.” CMS Email at 23 (Sept. 16, 2024). Yet Alignment did not contest this view, by asserting, for example, in its email (as it subsequently did in the district court) that the Protocols require more than making Spanish surveys available on request. It compounded this omission by failing to raise a Chenery objection in the district court when CMS responded with arguments in support of its interpretation of the Protocols.

Second, Alignment contends that CMS’s interpretation of the Protocols fails to “treat like cases alike.” Appellant’s Br. at 28 (citation omitted). By this account, the Protocols cannot permit vendors to vary how they 'undertake their datamatching processes' because such variance would permit “some vendors to take an ‘especially conservative’ approach to data handling while accepting that others will take more permissive approaches.” Reply Br. at 21–22. Alignment relies on Consolidated Edison Company of New York v. FERC , 45 F.4th 265 (D.C. Cir. 2022), where the court held that the decision to allocate the costs of two projects using a particular analytic method was arbitrary and capricious because the agency had used a different method for allocating the costs of a similar project without sufficient explanation. Id. at 278-81. That holding accords with recognizing that “an agency must provide an adequate explanation to justify treating similarly situated parties differently.” Comcast Corp. v. FCC , 526 F.3d 763, 769 (D.C. Cir. 2008); see, e.g., Baltimore Gas & Elec. Co. v. FERC , 954 F.3d 279, 286 (D.C. Cir. 2020); Grayscale Invs., LLC v. SEC , 82 F.4th 1239, 1245 (D.C. Cir. 2023). 'If a party plausibly alleges it has received inconsistent treatment under Page 15the same rule or standard, [the court] must consider whether the agency has offered a reasonable and coherent explanation for the seemingly inconsistent results.” Grayscale Invs., 82 F.4th at 1245 (citation omitted).

By contrast, Alignment has not shown that it was treated inconsistently, such as by pointing to when CMS has suppressed survey results in similar circumstances. Instead, Alignment offers a theoretical objection that permitting insurance plans and vendors to decide between themselves how to administer the survey in non-English languages could affect survey results. Appellant’s Br. at 28–29; Reply Br. at 21–22. Perhaps so, but Alignment fails to offer a basis on which the court could conclude this regulatory flexibility leads CMS to treat similarly situated cases differently or did so here.

Relatedly, Alignment turns to the private nondelegation doctrine, under which private entities must “function subordinately to the agency and . . . subject to its authority and surveillance.” FCC v. Consumers’ Research , 606 U.S. 656, 692 (2025). Alignment contends, third, that CMS’s interpretation of the Protocols violates this doctrine, pointing to CMS's statement that it 'does not get involved in how survey vendors implement language preference data.” Appellant's Br. at 18 (quoting CMS Email at 13 (Sept. 23, 2024)). Yet CMS’s interpretation does not confer “decisionmaking authority to outside entities.” Hight v. U.S. Dep’t of Homeland Sec. , 135 F.4th 996, 1009 (D.C. Cir. 2025) (quoting Louisiana Pub. Serv. Comm'n v. FERC , 860 F.3d 691, 696 (D.C. Cir. 2017) (ellipsis omitted)). While the regulatory scheme vests some discretion in insurance providers and vendors to decide how to collect data on a plan’s performance, CMS retains ultimate authority by overseeing the collection process and deciding whether the survey results are accurate Page 16enough to be included in scores for rating individual Medicare Advantage plans. Specifically,

CMS reviews the quality of the data on which performance, scoring and rating of a measure is based before using the data to score and rate performance or in calculating a Star Rating. This includes review of variation in scores among [Medicare Advantage] organizations . . . and the accuracy, reliability, and validity of measures and performance data before making a final determination about inclusion of measures in each year’s Star Ratings.

42 C.F.R. § 422.164(b). Under the plain text of the regulation, CMS decides whether the survey results are sufficiently accurate to be included in the Star Ratings calculation, and the vendor’s administration of the survey “can have only the legal (or, indeed, practical) effect” that CMS permits. Consumers’ Research , 606 U.S. at 694. The survey vendor serves a factgathering role here that raises no nondelegation issue. Southwest Airlines Co. v. Transportation Sec. Admin. , 650 F.3d 752, 757–58 (D.C. Cir. 2011).

Finally, no more persuasive is Alignment’s challenge that CMS's investigation or explanation was inadequate. See Appellant’s Br. at 35–37. After examining the survey results, CMS found “no evidence that there [was] an issue with the reliability of the [] Survey data.” CMS Email at 12 (Sept. 23, 2024). CMS concluded that the survey results indicated “unusually good access to Spanish-language surveys” for the two contracts at issue, CMS Email at 8 (Sept. 25, 2024), and offered several plausible explanations for the fluctuation in the data, including that some individuals who speak Spanish may have taken the survey in English, CMS Email at 16 (Sept. 19, 2024). CMS also contacted the survey vendor, which attested Page 17to following the relevant requirements. Id. Finding no support for Alignment's concern that the vendor had failed to make Spanish-language surveys available or that such a failure had rendered the survey results unreliable, CMS explained its conclusion during several weeks of email correspondence with Alignment. See supra Part I.B. Alignment, thus, fails to show that CMS acted unreasonably by failing to take further action. Moreover, CMS's interpretation of the Protocols would not leave Alignment without recourse in the future. Under the Protocols, Alignment has the option to request the vendor “to ‘double stuff’ mail packets to include both English and Spanish language surveys,' CMS Email at 23 (Sept. 16, 2024); Protocols at 50, seemingly obviating disputes about Spanishlanguage preference because everyone mailed a survey would receive the questionnaire in both English and Spanish.

Accordingly, the court affirms the grant of summary judgment in relevant part to CMS.