Medicare and Medicaid Programs; CY 2026 Payment Policies Under the Physician Fee Schedule and Other Changes to Part B Payment and Coverage Policies; Medicare Shared Savings Program Requirements; and Medicare Prescription Drug Inflation Rebate Program

Section 1848(c) of the Act requires that each service paid under the PFS be composed of three components: work, practice expense (PE), and malpractice (MP) expense. As required by section 1848(c)(2)(C)(iii) of the Act, beginning in CY 2000, MP RVUs are resource-based. Section 1848(c)(2)(B)(i) of the Act also requires that we review, and if necessary, adjust RVUs no less often than every 5 years. As explained in the CY 2011 PFS final rule with comment period (75 FR 73208), MP RVUs for new and revised codes effective before the next 5-year review of MP RVUs were determined either by a direct crosswalk from a similar source code or by a modified crosswalk to account for differences in work RVUs between the new/revised code and the source code. For the modified crosswalk approach, we adjusted (or scaled) the MP RVU for the new/revised code to reflect the difference in work RVU between the source code and the new/revised work RVU (or, if greater, the difference in the clinical labor portion of the fully implemented PE RVU) for the new code. For example, if the proposed work RVU for a revised code was 10 percent higher than the work RVU for its source code, the MP RVU for the revised code would be increased by 10 percent over the source code MP RVU. Under this approach, the same risk factor (RF) was applied for the new/revised code and source code, but the work RVU for the new/revised code was used to adjust the MP RVUs for risk.

We consider the following factors when we determine MP RVUs for individual PFS services: (1) specialty-level risk values derived from data on specialty-specific MP premiums incurred by practitioners; (2) service-level risk values derived from Medicare claims data of the weighted average risk values of the specialties that furnish each service; and (3) an intensity/complexity of service adjustment to the service-level risk value based on either the higher of the work RVU or clinical labor portion of the direct PE RVU. In the CY 2016 PFS final rule with comment period (80 FR 70906 through 70910), we discussed this methodology and finalized a policy to begin conducting annual MP RVU updates to reflect changes in the mix of practitioners providing services (using Medicare claims data), and to adjust MP RVUs for risk for intensity and complexity (using the work RVU or clinical labor RVU). We also finalized a policy to modify the specialty mix assignment methodology (for both MP and PE RVU calculations) to use an average of the three most recent years of data instead of a single year of data. Under this approach, for new and revised codes, we generally assign a specialty-level risk factor to individual codes based on the same utilization assumptions we make regarding the specialty mix we use for calculating PE RVUs and for PFS budget neutrality. We continue to use the work RVU or clinical labor RVU to adjust the MP RVU for each code for intensity and complexity. In finalizing this policy, we stated that the specialty-level risk factors would continue to be updated through notice and comment rulemaking every 5 years using updated premium data but would remain unchanged between the 5-year reviews.

In the CY 2018 PFS proposed rule (82 FR 33965 through 33970), we proposed to update the specialty-level risk factors used in the calculation of MP RVUs prior to the next required 5-year update (CY 2020) using the updated MP premium data that were used in the eighth Geographic Practice Cost Index (GPCI) update for CY 2017; however, the proposal was ultimately not finalized for CY 2018.

Section 1848(e)(1)(C) of the Act requires us to review, and if necessary, adjust the GPCIs at least every 3 years. In the CY 2020 PFS final rule (84 FR 62606 through 62615), we implemented the fourth review and update of MP RVUs, and we also conducted the statutorily required 3-year review of the GPCIs. The MP premium data used to update the MP GPCIs are the same data used to determine the specialty-level risk factors, which are used in the calculation of MP RVUs. Therefore, to increase efficiency, we finalized a policy to align the update of MP premium data and specialty-level risk factors with the update to the MP GPCIs. We finalized a policy to review, and if necessary, update the MP RVUs at least every 3 years, similar to our review and update of the GPCIs.

In the CY 2023 PFS final rule, we conducted the statutorily required review of the MP RVUs and GPCIs (87 FR 69634 through 69641). We refer to this review and update of the MP RVUs as the “CY 2023 update.” As part of this review, we finalized a methodological improvement to move from MP risk factors to a MP risk index. The risk index is calculated as a ratio of the specialty's national average premium to the volume-weighted national average premium across all specialties. We finalized this methodological improvement to increase consistency with the calculation of MP RVUs, so that changes in the MP risk index reflect changes in payment, as opposed to changes relative only to the specialty with the lowest national average premium. 2. Methodology for the Revision of Resource-Based Malpractice (MP) RVUs a. General Discussion

We calculated the MP RVUs that we proposed for CY 2026 using updated MP premium data obtained from state insurance rate filings. The methodology used to calculate the CY 2026 resource-based MP RVUs largely parallels the process used in the CY 2023 update with continued improvements to our data collection process. To calculate the MP RVUs, we obtain information on specialty-specific MP premiums that are linked to specific services, and using this information, we derive relative risk values for the various specialties that

furnish a particular service. Because MP premiums vary by state and specialty, we weigh the MP premium data geographically and by specialty. We calculated the MP RVUs we proposed using four data sources: data on MP insurance premium rates presumed to be in effect as of December 31, 2023; CY 2023 Medicare payment and utilization data; higher of the CY 2025 final work RVUs or the clinical labor portion of the direct PE RVUs; and CY 2025 GPCIs. We used the higher of the CY 2025 final work RVUs or clinical labor portion of the direct PE RVUs in our calculation to develop the CY 2026 proposed MP RVUs while maintaining overall PFS budget neutrality.

Similar to the CY 2023 update, we calculated the proposed MP RVUs using specialty-specific MP premium data because they represent the expense incurred by practitioners to obtain MP insurance as reported by insurers. For CY 2026, we obtained the most current MP insurance premium data available, reflecting rates with a presumed effective date of no later than December 31, 2023, from insurers with the largest market share in each state. We identified insurers with the largest market share using the National Association of Insurance Commissioners (NAIC) 2023 market share report. This annual report provides State- level market share for entities that provide premium liability insurance (PLI) in a state. Premium data was downloaded from the System for Electronic Rates & Forms Filing Access Interface (SERFF) (accessed from the NAIC website) for participating States. For non-SERFF States, data was downloaded from the State-specific website (if available online) or obtained directly from the State's alternate access to filings. For SERFF States and non-SERFF States with online access to filings, we used the 2023 market share report to select insurance companies. These market share filings were the most current data available during the data collection and acquisition process.

MP insurance premium data was collected from all 50 States and the District of Columbia. We made efforts to collect filings from Puerto Rico; however, no recent filings were submitted at the time of data collection, and therefore, we used filings from the previous update. Consistent with the CY 2023 MP RVU update, we did not collect filings for the other U.S. territories: American Samoa, Guam, Virgin Islands, or Northern Mariana Islands. We collected MP insurance premium data for coverage limits of $1 million/$3 million, mature, claims-made policies (policies covering claims made, rather than those covering losses occurring, during the policy term). A $1 million/$3 million liability limit policy means that the most that would be paid on any claim is $1 million and the most that the policy would pay for claims over the timeframe of the policy is $3 million. We made adjustments to the premium data to reflect mandatory surcharges for patient compensation funds (PCF, funds used to pay for any claim beyond the state's statutory amount, thereby limiting an individual physician's liability in cases of a large suit) in states where participation in such funds is mandatory.

In the CY 2020 PFS final rule (84 FR 62607 through 62610), we finalized methodological improvements that expanded the specialties and amount of filings data used to develop the proposed risk factors, which are used to develop the proposed MP RVUs. Premium data were included for all physician and nonphysician practitioner (NPP) specialties, and all risk classifications available in the collected rate filings. Although premium data were collected from all States, the District of Columbia, and previous filings for Puerto Rico were utilized, not all specialties had distinct premium data in the rate filings from all States. b. Methodological Refinements

For the CY 2026 update, we did not propose any major methodological refinements to the development of MP premium data. However, we have continued to refine the universe of specialties subject to imputation and sources of imputation for each specialty. For the CY 2023 update, premium data for the specialties of Geriatric Medicine, Hospitalist, Internal Medicine, Medical Oncology, Pain Management, and Preventive Medicine were augmented with some imputed data, but sufficient data was collected for these specialties during this CY 2026 update such that imputation was unnecessary. Additionally, Allergy/Immunology was previously used as the imputation source for both Osteopathic Manipulative Medicine and Addiction Medicine. For this CY 2026 update, more clinically similar specialties were used as the imputation source for these specialties. c. Steps for Calculating Malpractice RVUs

Calculation of the MP RVUs conceptually follows the specialty- weighted approach used in the CY 2015 PFS final rule with comment period (79 FR 67591), along with the methodological improvements established in the CY 2023 PFS final rule (87 FR 69634 through 69641). The specialty-weighted approach bases the MP RVUs for a given service on a weighted average of the risk index of all specialties furnishing the service. This approach ensures that all specialties furnishing a given service are reflected in the calculation of the MP RVUs. The steps for calculating the MP RVUs are described below.

Step (1): Compute a preliminary national average premium for each specialty.

Insurance rating area MP premiums for each specialty are mapped to the county level. The specialty premium for each county is then multiplied by its share of the total U.S. population (from the U.S. Census Bureau's 2018 to 2022 American Community Survey (ACS) 5-year estimates). This contrasts with the method used for creating national average premiums for each specialty in the 2015 update; in that update, specialty premiums were weighted by the total RVU per county, rather than by the county share of the total U.S. population. We refer readers to the CY 2016 PFS final rule with comment period (80 FR 70909) for a discussion of why we have adopted a weighting method based on share of total U.S. population. This calculation is then divided by the average MP GPCI across all counties for each specialty to yield a normalized national average premium for each specialty. The specialty premiums are normalized for geographic variation so that the locality cost differences (as reflected by the 2025 GPCIs) would not be counted twice. Without the geographic variation adjustment, the cost differences among fee schedule areas would be reflected once under the methodology used to calculate the MP RVUs and again when computing the service specific payment amount for a given fee schedule area.

Step (2): Determine which premium service risk groups to use within each specialty.

Some specialties had premium rates that differed for surgery, surgery with obstetrics, and non-surgery. These premium classes are designed to reflect differences in risk of professional liability and the cost of MP claims if they occur. To account for the presence of different classes in the MP premium data and the task of mapping these premiums to procedures, we calculated a distinct risk index for surgical, surgical with obstetrics, and nonsurgical procedures where applicable. However, the availability of data by surgery and non- surgery varied across specialties. Historically, no single approach accurately addressed the variability in

premium class among specialties, and we previously employed several methods for calculating average premiums by specialty.

Developing Distinct Service Risk Groups: We determined that there was sufficient data for surgery and non-surgery premiums, as well as sufficient differences in rates between classes for 17 specialties. These specialties are listed in Table A-M1. The CY 2026 update uses the same structure of specialty/service risk group as the CY 2023 update. For all other specialties (those that are not listed in Table A-M1) that typically do not distinguish premiums as previously described, a single risk index value was calculated, and that specialty risk index value was applied to all services performed by those specialties. [GRAPHIC] [TIFF OMITTED] TR05NO25.095

Step (3): Calculate a risk index for each specialty.

The relative differences in national average premiums between specialties are expressed in our methodology as a specialty-level risk index. These risk index values are calculated by dividing the national average premium for each specialty by the volume-weighted national average premium across all specialties. Risk index values less than one correspond to specialties with relatively lower malpractice risk than average, and values greater than one correspond to specialties with relatively higher malpractice risk. The volume-weighted national average premium was calculated as the sum of the product of the national average premium and total CY 2023 PE and work RVUs for each specialty/service risk group, then dividing by total CY 2023 PE and work RVUs across all specialties. (a) Technical Component (TC) Only Services

For the CY 2020 update of the MP RVUs (84 FR 62606 through 62615), we finalized that we would assign a risk factor of 1.00, which was the lowest physician specialty risk factor (allergy/immunology), to TC-only services due to a lack of sufficient professional liability premium data. For the proposed CY 2023 update of the MP RVUs (87 FR 46016), our expanded data collection efforts resulted in sufficient premium data such that we could directly assign a risk value for TC-only services without the need for mapping. However, due to a technical error, we continued to assign a 1.0 risk factor for all TC-only services which resulted in an incorrect calculation of the proposed MP RVUs for TC- only services. In the CY 2023 PFS final rule (87 FR 69641), we finalized a correction to this ratesetting error for the 2023 update of the MP RVUs that again mapped TC-only services to allergy/immunology, which had a risk index value of 0.430. We stated that using this risk value will correct the identified error, while also maintaining as much stability as possible for TC-only services so that there is not a major shift in value from current MP RVUs for the technical and professional components.

For this CY 2026 update of the MP RVUs, we are proposing to map TC- only services to the specialty allergy/immunology, which now has a risk index value of 0.427. Mapping the TC-only services to the specialty allergy/immunology would be consistent with the CY 2020 and 2023 updates of the MP RVUs and maintain stability in our ratesetting process. We requested comments regarding the risk index value for TC- only services. Table A-M2 shows the risk index values by specialty type and service risk group. [GRAPHIC] [TIFF OMITTED] TR05NO25.096

[GRAPHIC] [TIFF OMITTED] TR05NO25.097

[GRAPHIC] [TIFF OMITTED] TR05NO25.098

Step (4): Calculate MP RVUs for each CPT/HCPCS code.

Resource-based MP RVUs were calculated for each CPT/HCPCS code that has work or PE RVUs. The first step was to identify the percentage of services furnished by each specialty for each respective CPT/HCPCS code. This percentage was then multiplied by each respective specialty's risk index value as calculated in Step 3. The products for all specialties for the CPT/HCPCS code were then added together, yielding a specialty-weighted service specific risk index reflecting the weighted MP costs across all specialties furnishing that procedure. The service specific risk index was multiplied by the greater of

the work RVU or clinical labor portion of the direct PE RVU for that service, to reflect differences in the complexity and risk-of-service between services.

For low volume services codes, we finalized in the CY 2018 PFS final rule (82 FR 53000 through 53006) a proposal to apply the list of expected specialties instead of the claims-based specialty mix for low volume services to address stakeholder concerns about the year to year variability in PE and MP RVUs for low volume services (which also includes no volume services); these are defined as codes that have 100 allowed services or fewer. These service-level overrides are used to determine the specialty for low volume procedures for both PE and MP.

In the CY 2018 PFS final rule (82 FR 53000 through 53006), we also finalized our proposal to eliminate general use of an MP-specific specialty-mix crosswalk for new and revised codes. However, we indicated that we would continue to consider, in conjunction with annual recommendations, specific recommendations regarding specialty mix assignments for new and revised codes, particularly in cases where coding changes are expected to result in differential reporting of services by specialty, or where the new or revised code is expected to be low-volume. Absent such information, the specialty mix assumption for a new or revised code would derive from the analytic crosswalk in the first year, followed by the introduction of actual claims data, which is consistent with our approach for developing PE RVUs.

For CY 2026, we solicited public comment on the list of expected specialties. The list of codes and expected specialties is available on our website under downloads for the CY 2026 PFS final rule at http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeeSched/PFS-Federal-Regulation-Notices.html.

We received public comments on the list of expected specialties. The following is a summary of the comments we received and our responses.

Comment: Commenters recommended some additional CPT codes to be added to the CY 2026 PFS Proposed Rule Anticipated Specialty Assignment for Low Volume Services lists.\120\

\120\ https://www.cms.gov/files/zip/cy-2026-pfs-proposed-rule-anticipated-specialty-assignment-low-volume-services.zip.

Response: We appreciate commenters' suggested additions of low volume services CPT codes to the CY 2026 PFS Proposed Rule Anticipated Specialty Assignment for Low Volume Services list. We refer readers to the PE RVU Methodology section of this final rule for a discussion regarding the list and the suggested additions for CY 2026.

Step (5): Rescale for budget neutrality.

The statute requires that changes to fee schedule RVUs must be budget neutral. Thus, the last step is to adjust for relativity by rescaling the proposed MP RVUs so that the total proposed resource- based MP RVUs are equal to the total current resource-based MP RVUs scaled by the ratio of the pools of the proposed and current MP and work RVUs. This scaling is necessary to maintain the work RVUs for individual services from year to year while also maintaining the overall relationship among work, PE, and MP RVUs.

Specialties Excluded from Ratesetting Calculation: In section II.B. of this final rule, Determination of Practice Expense Relative Value Units, we discuss specialties that are excluded from ratesetting for the purposes of calculating PE RVUs. We proposed to treat those excluded specialties in a consistent manner for the purposes of calculating MP RVUs. We note that all specialties are included for purposes of calculating the final BN adjustment. The list of specialties excluded from the ratesetting calculation for the purpose of calculating the PE RVUs that we proposed to also exclude for the purpose of calculating MP RVUs is available in section II.B. of this final rule, Determination of Practice Expense Relative Value Units. The resource-based MP RVUs are shown in Addendum B, which is available on the CMS website under the downloads section of the CY 2026 PFS rule at https://www.cms.gov/medicare/payment/fee-schedules/physician/federal-regulation-notices.

Because a different share of the resources involved in furnishing PFS services is reflected in each of the three fee schedule components, implementation of the resource-based MP RVU update will have much smaller payment effects than implementing updates of resource-based work RVUs and resource-based PE RVUs. On average, work currently represents about 50.9 percent of payment for a service under the fee schedule, PE about 44.8 percent, and MP about 4.3 percent. Therefore, a 25 percent change in PE RVUs or work RVUs for a service would result in a change in payment of about 11 to 13 percent. In contrast, a corresponding 25 percent change in MP values for a service would yield a change in payment of only about 1 percent. Estimates of the effects on payment by specialty type are detailed in section VII. of this final rule, the Regulatory Impact Analysis.

Additional information on our methodology for updating the MP RVUs is available in the “Interim Report for the CY 2026 Update of GPCIs and MP RVUs for the Medicare Physician Fee Schedule,” which is available on the CMS website under the downloads section of the CY 2026 PFS final rule at https://www.cms.gov/medicare/payment/fee-schedules/physician/federal-regulation-notices.

We received public comments on these proposals. The following is a summary of the comments we received and our responses.

Comment: Commenters were supportive of the 2026 update to the risk index values used in the calculation of the MP RVUs. They commend CMS' continued work to expand data collection efforts by collecting as much specialty-specific data as possible to reflect the most accurate trends in malpractice premiums. Commenters requested that CMS continue to use more specialty-specific data not subject to imputation as well as increased transparency in the overall MP RVU update methodology.

Response: We appreciate the commenters for their feedback and support. We will continue to expand our data collection efforts for future updates of the Malpractice Risk Index. For an in-depth analysis and transparency of the MP RVU update process, we remind commenters of the interim and final reports referenced earlier in this section of the rule that outlines all steps of the update process and data imputation methodologies.

Comment: Commenters requested that CMS revisit their 2016 policy regarding exemptions for add-on codes and establish a minimum floor for MP RVUs where the practicing specialty's RVUs are significantly low. Commenters stated this would prevent a 0 MP RVU value for specialties that have a low, but still measurable, malpractice risk.

Response: For the CY 2026 PFS proposed rule, we did not make any proposals regarding our policy to implement an MP RVU floor of 0.01 with an exemption for add-on codes, as finalized in the CY 2016 PFS final rule (80 FR 70908). We direct readers to the CY 2016 PFS final rule (80 FR 70908) for further discussion on this exemption for add-on codes and we may consider modifications to this policy for future rulemaking.

Comment: A commenter suggested that CMS make changes to the specialty premium data source for a few specialties that they believe are incorrectly mapped for the purposes of

data imputation. The commenter also noted that they believe CMS could improve the imputation methodology by publishing impacts for all CMS specialties instead of mapping to related specialties in the regulatory impact table included in all PFS Federal Register notices. Additionally, the commenter stated that they would like CMS to work with the RUC to better identify appropriate crosswalks when necessary. The commenter's requested changes to mappings used in data imputation are as follows: 72-Pain Management to 11-Internal Medicine, 98- Gynecologist/oncologist to 91-Surgical oncology, C0-Sleep Medicine to 13-Neurology, and 85-Maxillofacial surgery (ALL) to 04-Otolaryngology (SURG).

Response: We appreciate the commenter's revised mapping suggestions for some specialties that require imputation of premium data. We also note that we will continue to work with all interested parties to improve the data used for calculating risk index values. We continue to believe that the list of CMS specialties contained in the regulatory impact table (Table 92 CY 2026 PFS Estimated Impact on Total Allowed Charges by Specialty) (90 FR 32803) is a useful tool to assist with mapping premium data when specialty-specific premium data are not included in a filing. However, as we previously discussed in this rule, we have adopted policies to improve our data imputation and employ partial imputation based on available data to approximate the premiums when we do not have complete specialty-specific premium data, as reflected in Table 8.C (Source Specialty/Service Risk Group for Imputation for Updated PLI Premium Data) of the interim report “CY2026 Medicare Physician Fee Schedule (PFS) Update to the Geographic Practice Cost Indices (GPCIs) and Malpractice (MP) Risk Index”. For the suggested mapping of 72-Pain Management to 11-Internal Medicine, we disagree with the commenters, because for the 2026 update of the MP risk index values for 72-Pain Management, we were able to collect sufficient premium data such that data imputation was not required and we were therefore able to use the actual premium data to formulate a risk index value for this specialty. For the suggested mappings of 98- Gynecologist/oncologist to 91-Surgical oncology and C0-Sleep Medicine to 13-Neurology, we also disagree with the commenter and believe this comment was an error. 98-Gynecologist/oncologist is already mapped to 91-Surgical oncology and C0-Sleep Medicine is already mapped to 13- Neurology. We made these mapping changes for the last update of the MP Risk Index values in the CY 2023 PFS final rule (87 FR 69641) and continued with the mappings for the CY 2026 update. Lastly, for the suggested mapping of 85-Maxillofacial surgery (ALL) to 04- Otolaryngology (SURG), after further consideration and review of the commenter's request, we are finalizing a change for these specialties for the purposes of data imputation to reflect that 85-Maxillofacial surgery (ALL) is now mapped to 04-Otolaryngology (SURG).

After consideration of public comments, we are finalizing the CY 2026 update as proposed with a modification to one specialty mapping used in data imputation as described previously.

Section 1848(e)(1)(A) of the Act requires us to develop separate Geographic Practice Cost Indices (GPCIs) to measure relative cost differences among localities compared to the national average for each of the three fee schedule components (that is, work, practice expense (PE), and malpractice (MP)). We discuss the localities established under the PFS later in this section. Although the statute requires that the PE and MP GPCIs reflect full relative cost differences, section 1848(e)(1)(A)(iii) of the Act requires that the work GPCIs reflect only one-quarter of the relative cost differences compared to the national average. In addition, section 1848(e)(1)(G) of the Act sets a permanent 1.5 work GPCI floor for services furnished in Alaska beginning January 1, 2009, and section 1848(e)(1)(I) of the Act sets a permanent 1.0 PE GPCI floor for services furnished in Frontier States (as defined in section 1848(e)(1)(I) of the Act) beginning January 1, 2011. Additionally, section 1848(e)(1)(E) of the Act provides for a 1.0 floor for the work GPCIs, which has been extended by many successive amendments to the statute. The 1.0 floor for the work GPCI under section 1848(e)(1)(E) of the Act was most recently extended by section 2206 of the Full-Year Continuing Appropriations and Extensions Act, 2025 (Pub. L. 119-4, enacted March 15, 2025) through September 30, 2025 (that is, for services furnished no later than September 30, 2025). Therefore, the proposed CY 2026 work GPCIs and summarized GAFs do not reflect the 1.0 work floor. Additionally, as required by sections 1848(e)(1)(G) and (I) of the Act, the 1.5 work GPCI floor for Alaska and the 1.0 PE GPCI floor for Frontier States are permanent, and therefore, are reflected in the CY 2026 proposed GPCIs.

Section 1848(e)(1)(C) of the Act requires us to review and, if necessary, adjust the GPCIs at least every 3 years. Section 1848(e)(1)(C) of the Act requires that, if more than 1 year has elapsed since the date of the last previous GPCI adjustment, the adjustment to be applied in the first year of the next adjustment shall be one-half of the adjustment that otherwise would be made. Therefore, since more than 1 year has passed since the previous GPCI update was implemented in CY 2023 and 2024, we proposed to phase in one-half of the proposed GPCI adjustment in CY 2026 and the remaining one-half of the adjustment for CY 2027.

We have completed our review of the GPCIs and are finalizing new GPCIs beginning for CY 2026 in this final rule. We also calculate a geographic adjustment factor (GAF) for each PFS locality. The GAFs are a weighted composite of each PFS locality's proposed work, PE, and MP GPCIs using the share of total RVUs that each component accounts for in the actual Medicare utilization from CY 2023. While we do not actually use GAFs in computing the PFS payment for a specific service, they are a useful metric for purposes of comparing overall costs and payments across fee schedule areas. The actual effect of GPCIs on payment for any actual service would deviate from the GAF to the extent that the proportions of work, PE and MP RVUs for the service differ from those reflected in the GAF.

See Addenda D and E to this proposed rule for the CY 2026 proposed GPCIs and summarized GAFs. These Addenda are available on the CMS website under the downloads section of the CY 2026 PFS final rule at https://www.cms.gov/medicare/payment/fee-schedules/physician/federal-regulation-notices. 2. Payment Locality Background

Prior to 1992, Medicare payments for physicians' services were made under the reasonable charge system. Payments under this system largely reflected the charging patterns of physicians, which resulted in large differences in payment for physicians' services among types of services, physician specialties and geographic payment areas.

Local Medicare carriers initially established 210 payment localities, to reflect local physician charging patterns and economic conditions. These localities changed little between the inception of Medicare in 1967 and the

beginning of the PFS in 1992. In 1994, we undertook a study that culminated in a comprehensive locality revision (based on locality resource cost differences as reflected by the GPCIs) that we implemented in 1997. The development of the current locality structure is described in detail in the CY 1997 PFS final rule (61 FR 34615) and the subsequent final rule with comment period (61 FR 59494). The revised locality structure reduced the number of localities from 210 to 89 and increased the number of Statewide localities from 22 to 34.

Section 220(h) of the Protecting Access to Medicare Act (PAMA) (Pub. L. 113-93, enacted April 1, 2014) required modifications to the payment localities in California for payment purposes beginning with 2017. As a result, in the CY 2017 PFS final rule (81 FR 80265 through 80268) we established 23 additional localities, increasing the total number of PFS localities from 89 to 112. Subsequently, we operationalized a technical refinement to retire several California localities that were no longer operationally necessary, resulting in a reduction of unique California localities from 32 to 29 from CY 2024 on. We refer readers to the discussion of this technical refinement in the CY 2023 (87 FR 69621 through 69625) and 2024 (88 FR 78985 through 78987) PFS final rules, and the section below. As a result, the current 109 payment localities include 34 Statewide areas (that is, only one locality for the entire State) and 72 localities in the other 16 States, with 10 States having two localities, two States having three localities, one State having four localities, and three States having five or more localities. The remainder of the 109 PFS payment localities are comprised as follows: the combined District of Columbia, Maryland, and Virginia suburbs; Puerto Rico; and the Virgin Islands. We noted that the localities generally represent a grouping of one or more constituent counties.

The current 109 fee schedule areas, also referred to as payment localities, are defined alternatively by State boundaries (Statewide areas for example, Wisconsin), metropolitan areas (for example, Metropolitan St. Louis, MO), portions of a metropolitan area (for example, Manhattan), or rest-of-state areas that exclude metropolitan areas (for example, Rest of Missouri). This locality configuration is used to calculate the GPCIs that are in turn used to calculate geographically adjusted payments for physicians' services under the PFS.

As stated in the CY 2011 PFS final rule with comment period (75 FR 73261), changes to the PFS locality structure would generally result in changes that are budget neutral within a State. For many years, before making any locality changes, we have sought consensus from among the professionals whose payments would be affected. We refer readers to the CY 2014 PFS final rule with comment period (78 FR 74384 through 74386) for further discussion regarding additional information about locality configuration considerations. 3. GPCI Update

As required by the statute, we developed GPCIs to measure relative cost differences among payment localities compared to the national average for each of the three fee schedule components (that is, work, PE, and MP). The changes to the proposed CY 2026 GPCIs for each locality reflect the updated resource cost data in each area to better adjust PFS payments for geographic cost differences compared to national average costs. We noted that the changes in the proposed GPCIs reflect the statutory floors and limitations on variation previously discussed that may advantage some rural localities. We describe the data sources and methodologies we use to calculate each of the three GPCIs later in this section. Additional information on the CY 2026 GPCI update is available in an interim report, “Interim Report for the CY 2026 Update of GPCIs and MP RVUs for the Medicare PFS,” on our website located under the downloads section for the CY 2026 PFS final rule at https://www.cms.gov/medicare/payment/fee-schedules/physician/federal-regulation-notices. a. Work GPCIs

The work GPCIs are designed to reflect the relative cost of physician labor by Medicare PFS locality. As required by statute, the work GPCI reflects one quarter of the relative wage differences for each locality compared to the national average.

To calculate the work GPCIs, we use wage data for nine professional specialty occupation categories, adjusted to reflect one-quarter of the relative cost differences for each locality compared to the national average, as a proxy for physicians' wages. Physicians' wages are not included in the occupation categories used in calculating the work GPCI because Medicare payments are a key determinant of physicians' earnings. Including physician wage data in calculating the work GPCIs would potentially introduce some circularity to the adjustment since Medicare payments typically contribute to or influence physician wages. That is, including physicians' wages in the physician work GPCIs would, in effect, make the indices, to some extent, dependent upon Medicare payments.

The work GPCI updates in CYs 2001, 2003, 2005, and 2008 were based on professional earnings data from the 2000 Census. However, for the CY 2011 GPCI update (75 FR 73252), the 2000 data were outdated, and wage and earnings data were not available from the more recent Census because the “long form” was discontinued. Therefore, we used the median hourly earnings from the 2006 through 2008 Bureau of Labor Statistics (BLS) Occupational Employment and Wage Statistics (OEWS), formerly known as Occupational Employee Statistics (OES), wage data as a replacement for the 2000 Census data. The BLS OEWS data meet several criteria that we consider to be important for selecting a data source for purposes of calculating the GPCIs. For example, the BLS OEWS wage and employment data are derived from a large sample size of approximately 200,000 establishments of varying sizes nationwide from every metropolitan area and can be easily accessible to the public at no cost. Additionally, the BLS OEWS is updated regularly, and includes a comprehensive set of occupations and industries (for example, 800 occupations in 450 industries). For the CY 2014 GPCI update, we used updated BLS OEWS data (2009 through 2011) as a replacement for the 2006 through 2008 data to compute the work GPCIs; for the CY 2017 GPCI update, we used updated BLS OEWS data (2011 through 2014) as a replacement for the 2009 through 2011 data to compute the work GPCIs; for the CY 2020 GPCI update, we used updated BLS OEWS data (2014 through 2017) as a replacement for the 2011 through 2014 data to compute the work GPCIs; and for the CY 2023 GPCI update, we used updated BLS OEWS data (2017 through 2020) as a replacement for the 2014 through 2017 data to compute the work GPCIs.

Because of its reliability, public availability, level of detail, and national scope, we believe the BLS OEWS data continue to be the most appropriate source of wage and employment data for use in calculating the work GPCIs (and as discussed later in this section, the employee wage component and purchased services component of the PE GPCI). Therefore, for the CY 2026 GPCI update, we used updated BLS OEWS data (2020 through 2023) as a replacement for the 2017 through 2020 data to compute the proposed work GPCIs.

The PE GPCIs are designed to measure the relative cost difference in the mix of goods and services comprising PEs (not including MP expenses) among the PFS localities as compared to the national average of these costs. Whereas the physician work GPCIs (and as discussed later in this section, the MP GPCIs) are comprised of a single index, the PE GPCIs are comprised of four component indices (employee wages; purchased services; office rent; and equipment, supplies and other miscellaneous expenses). The employee wage index component measures geographic variation in the cost of the kinds of skilled and unskilled labor that would be directly employed by a physician practice. Although the employee wage index adjusts for geographic variation in the cost of labor employed directly by physician practices, it does not account for geographic variation in the cost of services that typically would be purchased from other entities, such as law firms, accounting firms, information technology consultants, building service managers, or any other third-party vendor. The purchased services index component of the PE GPCI (which is a separate index from employee wages) measures geographic variation in the cost of contracted services that physician practices would typically buy. For more information on the development of the purchased service index, we refer readers to the CY 2012 PFS final rule with comment period (76 FR 73084 through 73085). The office rent index component of the PE GPCI measures relative geographic variation in the cost of typical physician office rents. For the medical equipment, supplies, and miscellaneous expenses component, we believe there is a national market for these items such that there is not significant geographic variation in costs. Therefore, the equipment, supplies and other miscellaneous expense cost index component of the PE GPCI is given a value of 1.000 for each PFS locality.

For the previous update to the GPCIs (implemented in CY 2023), we used 2017 through 2020 BLS OEWS data to calculate the employee wage and purchased services indices for the PE GPCI. As discussed previously in this section, because of its reliability, public availability, level of detail, and national scope, we continue to believe the BLS OEWS is the most appropriate data source for collecting wage and employment data. Therefore, in calculating the CY 2026 GPCI update, we used updated BLS OEWS data (2020 through 2023) as a replacement for the 2017 through 2020 data for purposes of calculating the employee wage component and purchased service index component of the PE GPCI. In calculating the CY 2026 GPCI update for the office rent index component of the PE GPCI, we used the 2018 through 2022 American Community Survey (ACS) 5-year estimates as a replacement for the 2015 through 2019 ACS data. c. Malpractice Expense (MP) GPCIs

The Malpractice Expense (MP) GPCIs measure the relative cost differences among PFS localities for the purchase of professional liability insurance (PLI). To ensure that premium data are homogenous and comparable across geographic areas, data were collected for policies with uniform coverage limits of $1 million per occurrence and $3 million aggregate ($1 million/$3 million). The MP GPCIs are calculated based on insurer rate filings of premium data for $1 million/$3 million mature claims-made policies (policies for claims made rather than losses occurring during the policy term). For the CY 2023 GPCI update, we used premium data presumed in effect as of December 31, 2020. The CY 2026 MP GPCI update reflects premium data presumed in effect no later than December 31, 2023. We noted that we finalized a few technical refinements to the MP GPCI methodology in CY 2017 and refer readers to the CY 2017 (81 FR 80270) PFS final rule for additional discussion of those. d. GPCI Cost Share Weights

For the CY 2026 GPCIs, we proposed to continue to use the current 2006-based MEI cost share weights for determining the proposed PE GPCI values. Specifically, we use the cost share weights to weight the four components of the PE GPCI: employee compensation, office rent, purchased services, and medical equipment, supplies, and other miscellaneous expenses, as shown in Table 31. We refer readers to the CY 2014 PFS final rule with comment period (78 FR 74382 through 74383), for further discussion regarding the 2006-based MEI cost share weights revised in CY 2014 that we also finalized for use in the CY 2017, CY 2020, and CY 2023 GPCI updates.

We note that we proposed and finalized to rebase and revise the MEI cost share weights for CY 2023, and we refer readers to the detailed discussion in section II.M. of the CY 2023 PFS final rule (87 FR 69688 through 69710). Due to the concurrent rebasing and revision of the MEI cost share weights during the CY 2023 GPCI update, we proposed and finalized to maintain the use of the 2006-based MEI cost share weights for the CY 2023 GPCIs, thus delaying the implementation of the rebased and revised 2017-based MEI cost share weights for this purpose. We refer readers to our discussion about using the rebased and revised MEI cost share weights for purposes of proportioning the work, PE, and MP RVU pools in PFS ratesetting and for the purposes of updating the GPCIs in the CY 2023 PFS final rule (87 FR 69414 through 69415, 69619 through 69620, and 70212 through 70218). In those sections, we discussed our considerations for updating the MEI cost share weights for the RVUs and the GPCIs and the potential redistributive impact that making such a change would have had on PFS payments. We have historically updated the GPCI cost share weights to make them consistent with the most recent update to the MEI, which was most recently done for CY 2023. However, in light of the overall impacts of making this change and in the interest of maintaining stability in payments, we proposed and finalized to maintain the use of the currently used 2006-based MEI cost share weights for the CY 2023 final PE GPCIs. For the CY 2026 GPCI update, we have the same concerns about the potential redistributive effects that implementing the 2017-based MEI would have on PFS payments. Additionally, we have received data from the American Medical Association's (AMA) Physician Practice Information \121\ (PPI) and Clinician Practice Information \122\ (CPI) Surveys, however, these data lack the specific breakdown of practice expense that we would need to consider its use to weight the four components of the PE GPCI for CY 2026, including Office Rent and Purchased Services, which are reported in an aggregate buckets of general overhead costs and other expenses in the survey data. We refer readers to section VII. of this final rule for more discussion regarding a possible derivation of cost share weights for use in the PE GPCI from the PPI and CPI Survey.

\121\ https://www.ama-assn.org/system/files/table-1-results-from-ppi.pdf.

\122\ https://www.ama-assn.org/system/files/table-1-results-from-cpi-final.pdf.

We also note that maintaining the 2006-based MEI cost share weights for the CY 2026 GPCI update preserves consistency in the data used to update both the GPCI and PFS ratesetting inputs for CY 2026. We refer readers to section VII. of this final rule for additional discussion on this issue and the estimated impacts as it relates to PFS ratesetting and the GPCI update for

CY 2026. We also refer readers to the discussion regarding the PPI and CPI survey data in section II.B. of this final rule. In addition, we direct readers to the CY 2011 PFS final rule (75 FR 73256) where we similarly delayed implementation of updated MEI cost share weights in response to commenters' concerns about our separate, ongoing analysis that would inform future GPCI changes and the reallocation of labor- related costs from the medical equipment and supplies and miscellaneous component to the employee compensation component of the PE GPCI.

In the CY 2011 PFS final rule (75 FR 73256), we acknowledged that we typically update the GPCI cost share weights concurrently with the most recent MEI rebasing and revision, but in consideration of the commenters' concerns in response to the proposed rule, we did not use the revised cost share weights for the CY 2011 GPCIs and instead finalized the implementation of the rebased and revised MEI cost share weights through subsequent rulemaking. We sought comments on the 2017- based MEI cost share weights and the weights based on PPI and CPI Survey data for purposes of alternatives considered for the CY 2026 GPCIs and PFS ratesetting, given the estimated impacts discussed in section VII. of this final rule. We also sought comments on how best to proceed with implementation of the 2017-based MEI cost share weights or PPI and CPI Survey weights in the future. More specifically, we sought comment on how best to incorporate updated cost share weights into the PE GPCI if we were to implement them outside the statutorily required triennial update in which we phase in all aspects of the GPCI update through the previously discussed 2-year (one-half in each year) phase- in required by section 1848(e)(1)(C) of the Act. Section 1848(e)(1)(C) of the Act requires that, if more than 1 year has elapsed since the date of the last GPCI adjustment, the adjustment to be applied in the first year of the next adjustment shall be one-half of the adjustment that otherwise would be made. Therefore, we sought comment on potentially incorporating the updated cost share weights into the CY 2027 GPCIs. We note that we would not be required by statute to phase in the adjustment over 2 years as specified in section 1848(e)(1)(C) of the Act because, in CY 2027, no more than 1 year would have elapsed since this CY 2026 GPCI adjustment. Therefore, we also sought comment on whether it would be appropriate to use a multi-year transition to incorporate updated cost share weights for purposes of the PE GPCI and PFS ratesetting as we have done in the past when incorporating other new data into the PFS payment methodology (for example, the clinical labor update), or if, because updated cost share weights only impact the composition of the PE GPCI, such a transition would not be warranted. If we were to instead apply updated cost share weights for purposes of the PE GPCI and PFS ratesetting for CY 2028 or a later calendar year, we would be required under section 1848(e)(1)(C) of the Act to phase in the GPCI adjustments over 2 years. We sought comments on whether, in that case, it would be appropriate to similarly apply a transition to implement updated cost share weights for purposes of PFS ratesetting as well, and refer readers to section II.B and VII. of this final rule for more discussion regarding the alternatives considered and impacts of a phase-in of updated cost share weights in PFS ratesetting. The proposed CY 2026 GPCI cost share weights are displayed in Table AN-1. We note that the 2017-based MEI cost share weights as finalized in section II.M. of the CY 2023 PFS final rule (87 FR 69688 through 69708) are also displayed in Table AN-1 for awareness regarding potential future rulemaking and GPCI updates. As previously discussed, the PPI and CPI Survey data lack the specific breakdown of practice expense that we would need to consider its use to weight the four components of the PE GPCI for CY 2026, therefore, we refer readers to section VII. of this final rule for more discussion regarding a possible derivation of cost share weights for use in the PE GPCI from the PPI and CPI Survey for awareness regarding potential future rulemaking and GPCI updates. [GRAPHIC] [TIFF OMITTED] TR05NO25.099

Section 10324(c) of the Affordable Care Act added a new subparagraph (I) under section 1848(e)(1) of the Act to establish a 1.0 PE GPCI floor for physicians' services furnished in Frontier States effective January 1, 2011. In accordance with section 1848(e)(1)(I) of the Act, beginning in CY 2011, we applied a 1.0 PE GPCI floor for physicians' services furnished in States determined to be Frontier States. In general, a Frontier State is one in which at least 50 percent of the counties are “frontier counties,” which are those that have a population per square mile of less than 6. For more information on the criteria used to define a Frontier State, we refer readers to the FY 2011 Hospital Inpatient Prospective Payment System (IPPS)/Long- term Care Hospital PPS final rule (75 FR 50160 through 50161). There are no changes in the states identified as Frontier States for the CY 2026 PFS proposed rule. The qualifying states are: Montana; Wyoming; North Dakota; South Dakota; and Nevada. In

accordance with statute, we will apply a 1.0 PE GPCI floor for these states in CY 2026. f. Methodology for Calculating GPCIs in the U.S. Territories

Prior to CY 2017, for all the island territories other than Puerto Rico, the lack of comprehensive data about unique costs for island territories had minimal impact on GPCIs because we used either the Hawaii GPCIs (for the Pacific territories: Guam; American Samoa; and Northern Mariana Islands) or used the unadjusted national averages (for the Virgin Islands). In an effort to provide greater consistency in the calculation of GPCIs given the lack of comprehensive data regarding the validity of applying the proxy data used in the States in accurately accounting for variability of costs for these island territories, in the CY 2017 PFS final rule (81 FR 80268 through 80270), we finalized a policy to treat the Caribbean Island territories (the Virgin Islands and Puerto Rico) in a consistent manner. We do so by assigning the national average of 1.0 to each GPCI index for both Puerto Rico and the Virgin Islands. We refer readers to the CY 2017 PFS final rule for a comprehensive discussion of this policy. g. California Update to the Fee Schedule Areas Used for Payment Under Section 220(h) of the Protecting Access to Medicare Act

Section 220(h) of the PAMA added a new section 1848(e)(6) to the Act that modified the fee schedule areas used for payment purposes in California beginning in CY 2017. Prior to CY 2017, the fee schedule areas used for payment in California were based on the revised locality structure that was implemented in 1997 as previously discussed. Beginning in CY 2017, section 1848(e)(6)(A)(i) of the Act required that the fee schedule areas used for payment in California must be Metropolitan Statistical Areas (MSAs) as defined by the Office of Management and Budget (OMB) as of December 31 of the previous year; and section 1848(e)(6)(A)(ii) of the Act required that all areas not located in an MSA must be treated as a single rest-of-state fee schedule area. The resulting modifications to California's locality structure increased its number of fee schedule areas from 9 under the current locality structure to 27 under the MSA-based locality structure; although for the purposes of payment, the actual number of fee schedule areas under the MSA-based locality structure is 32. We refer readers to the CY 2017 PFS final rule (81 FR 80267) for a detailed discussion of this operational decision.

Section 1848(e)(6)(D) of the Act defined transition areas as the counties in fee schedule areas for 2013 that were in the rest-of-state locality, and locality 3, which was comprised of Marin County, Napa County, and Solano County. Section 1848(e)(6)(B) of the Act specified that the GPCI values used for payment in a transition area are to be phased in over 6 years, from 2017 through 2022, using a weighted sum of the GPCIs calculated under the new MSA-based locality structure and the GPCIs calculated under the PFS locality structure that was in place prior to CY 2017. That is, the GPCI values applicable for these areas during this transition period were a blend of what the GPCI values would have been for California under the locality structure that was in place prior to CY 2017, and what the GPCI values would be for California under the MSA-based locality structure. For example, in CY 2020, which represented the fourth year of the transition period, the applicable GPCI values for counties that were previously in the rest- of-state locality or locality 3 and are now in MSAs were a blend of two-thirds of the GPCI value calculated for the year under the MSA- based locality structure, and one-third of the GPCI value calculated for the year under the locality structure that was in place prior to CY 2017. The proportions continued to shift by one-sixth in each subsequent year so that, by CY 2021, the applicable GPCI values for counties within transition areas were a blend of five-sixths of the GPCI value for the year under the MSA-based locality structure, and one-sixth of the GPCI value for the year under the locality structure that was in place prior to CY 2017. Beginning in CY 2022, the applicable GPCI values for counties in transition areas were the values calculated solely under the new MSA-based locality structure; therefore, the phase-in for transition areas is complete. Additionally, section 1848(e)(6)(C) of the Act establishes a hold harmless requirement for transition areas beginning with CY 2017; whereby, the applicable GPCI values for a year under the new MSA-based locality structure may not be less than what they would have been for the year under the locality structure that was in place prior to CY 2017. There are 58 counties in California, 50 of which were in transition areas as defined in section 1848(e)(6)(D) of the Act. The eight counties that were not within transition areas are: Orange; Los Angeles; Alameda; Contra Costa; San Francisco; San Mateo; Santa Clara; and Ventura counties. We note that while the phase-in for transition areas is no longer applicable, the hold-harmless requirement is not time-limited, and therefore, is still in effect.

For the purposes of calculating budget neutrality and consistent with the PFS budget neutrality requirements as specified under section 1848(c)(2)(B)(ii)(II) of the Act, we finalized the policy to start by calculating the national GPCIs as if the fee schedule areas that were in place prior to CY 2017 are still applicable nationwide; then, for the purposes of payment in California, we override the GPCI values with the values that are applicable for California consistent with the requirements of section 1848(e)(6) of the Act. This approach to applying the hold harmless requirement is consistent with the implementation of the GPCI floor provisions that have previously been implemented--that is, as an after-the-fact adjustment that is made for purposes of payment after both the GPCIs and PFS budget neutrality have already been calculated.

Additionally, section 1848(e)(1)(C) of the Act requires that, if more than 1 year has elapsed since the date of the last GPCI adjustment, the adjustment to be applied in the first year of the next adjustment shall be one-half of the adjustment that otherwise would be made. For a comprehensive discussion of this provision, transition areas, and operational considerations, we refer readers to the CY 2017 PFS final rule (81 FR 80265 through 80268).

In the CY 2020 final rule (84 FR 62622), a commenter indicated that some of the distinct fee schedule areas that were used during the period between CY 2017 and CY 2018 are no longer necessary. Specifically, with regard to the Los Angeles-Long Beach-Anaheim MSA, which contains 2 counties (across two former unique locality numbers, 18 and 26) that are not transition areas, we acknowledge that we only needed more than one unique locality number for that MSA for payment purposes in CY 2017, which was the first year of the implementation of the MSA-based payment locality structure. Neither of the counties in the Los Angeles-Long Beach-Anaheim MSA (Orange County and Los Angeles County) are transition areas under section 1848(e)(6)(D) of the Act. Therefore, the counties were not subject to the aforementioned GPCI value incremental phase-in (which is no longer applicable) or the hold- harmless provision at section 1848(e)(6)(C) of the Act. Similarly, the San Francisco-Oakland-Berkeley MSA contains four

counties--San Francisco, San Mateo, Alameda, and Contra Costa counties--across three former unique locality numbers, 05, 06, and 07. These counties are not transition areas and will receive the same GPCI values, for payment purposes, going forward. In response to the comment, we acknowledged that we did not propose any changes to the number of fee schedule areas in California, but would consider the feasibility of a technical refinement to consolidate into fewer unique locality numbers; and if we determined that consolidation was operationally feasible, we would propose the technical refinement in future rulemaking. In light of the foregoing, for CY 2023, we proposed and finalized to identify the Los Angeles-Long Beach-Anaheim MSA, containing Orange County and Los Angeles County, by one unique locality number, 18, as opposed to two, thus retiring locality number 26, as it is no longer needed. Similarly, we proposed and finalized to identify the San Francisco-Oakland-Berkeley MSA containing San Francisco, San Mateo, Alameda, and Contra Costa counties by one unique locality number, 05, as opposed to four, thus retiring locality numbers 06 and 07, as they are no longer needed. Additionally, we noted that we would modify the MSA names as follows: the San Francisco-Oakland-Berkeley (San Francisco Cnty) locality (locality 05) would become San Francisco- Oakland-Berkeley (San Francisco/San Mateo/Alameda/Contra Costa Cnty), and Los Angeles-Long Beach-Anaheim (Los Angeles Cnty) locality (locality 18) would become Los Angeles-Long Beach-Anaheim (Los Angeles/ Orange Cnty). The refinement finalized in the CY 2024 PFS final rule (88 FR 78985 through 78987) ultimately changed the number of distinct fee schedule areas for payment purposes in California from 32 to 29. We noted that because Marin County is in a transition area and subject to the hold harmless provision at section 1848(e)(6)(C) of the Act, we needed to retain a unique locality number for San Francisco-Oakland- Berkeley (Marin Cnty), locality 52. We noted that these changes do not have any payment implications under the PFS. h. Alternatives Considered Related to List of Occupation Codes Used in the Work GPCI Calculation

As explained in the Work GPCIs section above, we utilize a refined list of occupation groups and codes from the Bureau of Labor Statistics (BLS) Occupational Employment and Wage Statistics (OEWS) data to calculate the work GPCI. Because of its reliability, public availability, level of detail, and national scope, we believe the BLS OEWS data continue to be the most appropriate source of wage and employment data for use in calculating the work GPCIs. For the CY 2023 GPCI update, we reviewed the occupation codes and groups used to capture geographic variation in professional wages to assess other potential codes and groups that could be used in addition to the current selections to calculate the work GPCI, with significant consideration given to the extent to which the data exist in the file (data existence) and how well the occupation codes are represented in the data (data sufficiency). Based on our review and commenters' response to the proposals, we finalized the addition of two new occupation groups (and their corresponding occupation codes), Management Occupations and Business and Financial Operation Occupations, to the preexisting seven occupation groups, and four new occupation codes to the pre-existing Computer, Mathematical, Life, and Physical Science group, and three occupation codes to the pre-existing Social Science, Community and Social Service, and Legal group in the CY 2023 PFS final rule (87 FR 69621 through 69625). The practical effect of the addition of these occupation groups and codes on the work GPCI was minimal because the statute at section 1848(e)(1)(A)(iii) of the Act requires that the work GPCI reflect only one quarter of cost differences, but their inclusion added meaningful data regarding the geographic variation in professional wages for CY 2023.

In the CY 2023 PFS final rule (87 FR 69631), some commenters stated that our methodologic changes to the work GPCI occupation groups and codes create unnecessary complexity and limited transparency. The commenters stated that CMS did not provide an impact analysis or criteria for inclusion (that is, how well it correlated as a proxy) other than significant consideration to the extent to which the data exists in the file (data existence) and how well the occupation codes are represented in the data (data sufficiency). The commenters stated that, without further explanation, two additional occupation groups were added to the previous seven occupation groups, which increased the greater than 100 current occupation codes by 60. A commenter believed that it is unlikely that the cumulation of so many professions will accurately reflect the relative difference in work of a single profession such as a physician; the commenter stated that, if one were to compare the BLS OEWS data file used for the work GPCI with that of the healthcare provider dataset, there is a discordance. The commenters agreed that the healthcare provider dataset should not be used for developing the work GPCI due to circularity, but believe it could be used to validate the proposed work GPCIs and to identify a much smaller subset of professions that would act as more reliable proxies than what was proposed. The commenters urged CMS to apply a smaller number of professions to the work GPCI, as they thought that doing so would result in a more reliable and accurate proxy for physician work, and provide more information about the correlation between physician work and the proxy professions to allow the public to verify its accuracy.

In response to commenters, we noted that we do not claim the proxy professions themselves, or the absolute wages of the proxy professionals are correlated to physician wages, but rather, that the geographic variation in proxy professional wages is similar to the geographic variation in physician wages.

We believed that there would be similar geographic variation if one were to compare the BLS OEWS data used for the work GPCI with data from a healthcare provider dataset. We continue to believe in the majority of instances, the earnings of physicians will vary among areas to the same degree that the earnings of other professionals across an array of industries vary. Further, we welcomed opportunities to discuss data sources that can be used to validate the work GPCI, similar to the analysis that we performed for residential and commercial rent data used for the office rent index for CY 2023.

For CY 2026, we analyzed the potential effect of using a consolidated set of occupation codes on the work GPCI and compared that effect to changes in work GPCI values that would occur utilizing the standard set of occupation codes, as finalized for CY 2023. We acknowledge that the use of a more parsimonious set of occupations could be an improvement if it results in essentially the same work GPCI values with increased simplicity and clarity for interested parties. We explored approaches to condense the list of occupation codes used in a more systematic manner, with the establishment of inclusion criteria for an occupation code such as level of education attainment and data completeness. For our analysis, we identified 274, 157 and 90 occupation codes with at least 50 percent, 75

percent, and 90 percent having a Bachelor's Degree or higher, excluding occupation codes in Group 29 that are paid on the Fee Schedule, respectively from the May 2023 OEWS data. We then applied various data completeness criteria thresholds to these occupation codes with wage data for at least 50 percent, 75 percent, and 90 percent of U.S. counties, resulting in the number of occupation codes displayed in Table AN-2. [GRAPHIC] [TIFF OMITTED] TR05NO25.100

Of these scenarios with various thresholds of the education attainment and data completeness inclusion criterion, we investigated two scenarios compared to the standard CY 2026 GPCI: (1) occupation codes with at least 75 percent of Bachelor's Degree or Higher excluding Group 29 and wage data for at least 50 percent of U.S. counties, resulting in a list of 57 occupation codes; and (2) occupation codes with at least 75 percent of Bachelor's Degree or Higher excluding Group 29 and wage data for at least 75 percent of U.S. counties, resulting in a list of 31 occupation codes from the May 2023 OWES data. Under these two scenarios, the work GPCIs result in changes relative to current CY 2025 work GPCI values that are nearly identical to those under the standard CY 2026 GPCI update, as shown in Table AN-3. [GRAPHIC] [TIFF OMITTED] TR05NO25.101

Based on the two scenarios' changes relative to current CY 2025 work GPCI values that are nearly identical to those under the standard CY 2026 GPCI update, we sought comment on the potential to establish clear inclusion criteria for occupation codes for the calculation of the work GPCI in future GPCI updates. We note that a smaller, standardized list of occupation codes that meet rigorous and clearly established thresholds for education attainment and data completeness would aid transparency in the work GPCI and be responsive to the commenters' requests.

Similar to the finalized addition of occupation groups and codes for the CY 2023 GPCI update, the practical effect of limiting the occupation groups and codes on the work GPCI would be minimal because the statute at section

1848(e)(1)(A)(iii) of the Act requires that the work GPCI reflect only one quarter of cost differences, but the limitation could aid transparency and allow for a greater degree of precision when tracking changes in geographic variation in professional wages across GPCI update years. i. GPCI Update Summary

As explained in the Background section above, section 1848(e)(1)(C) of the Act mandates the periodic review and adjustment of GPCIs. For each periodic review and adjustment, we published the proposed GPCIs in the PFS proposed rule to provide an opportunity for public notice and comment and allow us to consider whether any revisions in response to comments are warranted prior to implementation. The proposed CY 2026 updated GPCIs that we proposed for the first and second year of the 2- year phase-in, along with the GAFs, are displayed in Addenda D and E to this proposed rule available on our website under the supporting documents section of the CY 2026 PFS final rule web page at https://www.cms.gov/medicare/payment/fee-schedules/physician/federal-regulation-notices.

We noted that in recent GPCI updates, commenters have stated that there is a lack of transparency into the GPCI data and methodology used to derive the GPCIs. In response to the CY 2023 PFS proposed rule, a commenter stated that they cannot accurately validate CMS' GPCI calculations because there is little transparency and access to the data and methods used. The commenter stated that they submitted a comment on the CY 2022 PFS proposed rule urging CMS to provide more transparency into the GPCI calculations in general, including a more detailed description of the step-by-step methodology and the specific data files used to derive the GPCIs. In addition to making the RVUs by county available, the commenters also suggested CMS to make available the source data for the work GPCI by county, the source data for each component of the practice expense GPCI, and all budget neutrality adjustments and calculations.

The commenters stated that CMS provided these data prior to 2020 and that they used it to reproduce and validate the CMS methodology for calculating the GPCIs each year.

In the CY 2023 PFS final rule, in response to these comments, we referred readers to the step-by-step instructions provided in the final report, “Final Report for the CY 2023 Update of GPCIs and MP RVUs for the Medicare PFS,” on our website located under the supporting documents section for the CY 2023 PFS final rule at https://www.cms.gov/medicare/payment/fee-schedules/physician/federal-regulation-notices. We also referred readers to Table 4.A.1: Summary of Elements Required for GPCI Calculation in the final report, and the previous discussion, for the data sources used for the work GPCI and each component of the practice expense GPCI. As noted in the proposed and final rules for each GPCI update, we discuss the years and timeframes of data used from each source. We note that we provide web links to the publicly-available data sources used in the GPCI updates, the methodological parameters, as well as an overview of how we develop each GPCI component in the interim and final reports published with each proposed and final rule containing a GPCI update. This practice is consistent with previous updates. We also note that the budget neutrality adjustment and statutory floors applied after the budget neutrality adjustment are detailed in the note, “CY 2023 GPCI Update Note_County_Data,” on our website located under the downloads section for the CY 2023 PFS proposed and final rules at https://www.cms.gov/medicare/payment/fee-schedules/physician/federal-regulation-notices. We also reminded commenters that, in response to the commenters' concerns expressed in rulemaking for the CY 2020 GPCI update, we included more detailed steps in the final report, “Final Report for the CY 2020 Update of GPCIs and MP RVUs for the Medicare Phys Fee Sched_v19Feb2020”, which is available on the CMS website under the downloads section of the CY 2020 PFS final rule to assist interested parties in navigating these data. Additionally, as part of our ongoing commitment to transparency, we post the county-level data that we use to develop the proposed GPCIs, which allows interested parties to further examine and replicate our GPCI methodology. This file is also available on the CMS website under the Downloads section for the PFS, titled “CY 2023 Proposed Rule GPCI County-Level Data File.” We believe that we sufficiently addressed previous commenters' concerns for the CY 2023 GPCI update in the proposed and final rules and aforementioned CY 2020 and CY 2023 interim and final reports, but we sought comments related to any additional information specific to what data was provided prior to 2020 that is no longer provided. Based on a comparison of data and information in the interim and final reports, as well as the data file downloads, we have not identified any information or data that we have discontinued since 2020, as commenters have claimed. We sought feedback related to specific information and data that would aid transparency in a GPCI update.

We received public comments on these proposals. The following is a summary of the comments we received and our responses.

Comment: Several commenters expressed concerns over the expiring 1.0 GPCI work floor. Some of the commenters stated that if the 1.0 GPCI work floor is not extended, this would result in negative impacts, especially in rural and underserved areas. A commenter also stated that CMS should apply any claim adjustment automatically should there be a delay in enactment of an extension of the GPCI work floor policy, with a retroactive implementation date.

Response: The 1.0 work GPCI floor is established by statute and expired on September 30, 2025 (NOTE: If necessary, this date may be changed after the next round of clearance). CMS does not have the authority to extend the 1.0 work floor beyond the September 30, 2025 expiration and will process claims in accordance with statutory and regulatory requirements.

Comment: Several commenters opposed CMS' proposed CY 2026 GPCI decreases for Arkansas. Several of these commenters requested that CMS maintain the work floor of 1.0 and freeze Arkansas' 2025 GPCI values while a comprehensive review is conducted, or recalculate the Arkansas PE GPCI using current, disaggregated data that reflects real regional cost variation within the state, or at a minimum appropriately weighs the rate based on the population centers experiencing the highest costs, which service the majority of the state's population. The commenters also stated that the current GPCI methodology ignores regional variations within Arkansas, particularly in metropolitan areas like Washington, Benton, and Pulaski Counties. A commenter requested that we establish a timeline for evaluating Medicare localities to align with Core Based Statistical Areas (CBSA) designations, to automate county-level locality splits (that is, split any country >10 percent above statewide PE input cost for 2 consecutive years), and to meet with interested parties to collaborate on solutions. The commenter also requested that we update data annually to shorten the lag for fast-growing regions, and to apply the frontier-state PE floor to HRSA- designated Health Provider Shortage Areas (HPSAs). The

commenter requested that we recognize geographic variation in the supplies and equipment component of the PE GPCI and use commercial medical office rent surveys (that is, CoStar, CBRE) instead of ACS rent data.

Some of the commenters referenced the Arkansas House Concurrent Resolution 1007 (HCR1007) and stated that the Arkansas General Assembly adopted HCR1007 during the 2025 Regular Session and urged CMS to reevaluate Arkansas' GPCI treatment and create distinct PFS localities for Little Rock-North Little Rock-Conway (CBSA 30780) and Fayetteville- Springdale-Rogers (CBSA 2220). Some of the commenters also stated that the undervaluation of Arkansas' GPCI makes it challenging to recruit and retain physicians, particularly in rural and underserved areas. While a commenter recommended that CMS recognize Central Arkansas as a distinct locality to better align payment with actual costs, another commenter opposed regional Arkansas evaluation, arguing it would strain rural providers and exacerbate workforce challenges.

Response: We appreciate the commenters' feedback. However, we reiterate that the GPCIs are based on nationally-representative and publicly-available wage data from the BLS OEWS for the work GPCI and employee wage and purchased services components of the PE GPCI, and the Census Bureau's ACS data for the rent index component of the PE GPCI. We reiterate that the GPCIs are not an absolute measure of practices costs. Rather, they are a measure of relative resource cost differences among localities compared to the national average as informed by the data and are not intended to represent or measure changes in GDP or migration. We also remind the commenters that section 1848(e)(1)(I) of the Act defines Frontier State and sets the permanent PE GPCI floor of 1.0, therefore CMS does not have the authority to apply the 1.0 Frontier State PE GPCI floor to HPSAs. Additionally, section 1848(e)(1)(C) of the Act requires us to review, and if necessary, adjust the GPCIs at least every 3 years, therefore CMS does not have the authority to freeze the CY 2025 GPCIs. We note that the CY 2026 GPCI update was calculated using current, disaggregated data that reflects regional cost variation within each state, as requested by the commenter, and therefore the published CY 2026 PE GPCI in Addendum E does account for county-level differences weighted by Medicare PFS RVUs and population by country (as described in the “Interim Report for the CY 2026 Update of GPCIs and MP RVUs for the Medicare Physician Fee Schedule,” available on the CMS website under the downloads section of the CY 2026 PFS proposed rule at https://www.cms.gov/medicare/payment/fee-schedules/physician/federal-regulation-notices).

We appreciate the suggestions for revisions to the PFS locality structure. We did not propose any changes to the PFS locality structure and therefore will not finalize any changes to the PFS locality structure for CY 2026. However, we may consider changes to the PFS locality structure through future rulemaking.

Additionally, we would like to reiterate that the current PFS locality structure was developed and implemented in 1997 with minor modifications over the years. We have also considered more comprehensive changes to locality configurations. In 2008, we issued a draft comprehensive report detailing four different locality configuration options (www.cms.gov/physicianfeesched/downloads/ReviewOfAltGPCIs.pdf) and we refer readers to the CY 2010 PFS proposed rule (74 FR 33534) and subsequent final rule with comment period (74 FR 61757) for a detailed discussion of the public comments on the contractor's 2008 draft report detailing different locality configurations. All four of the potential alternative payment locality configurations reviewed in the report would increase the number of localities and separate higher cost, typically urban areas from lower cost, typically rural “Rest of State” areas. In general, payments to urban areas would increase while rural areas would see a decrease in payment under each of the options studied because they would no longer be grouped with higher cost “urbanized” areas. Disaggregation of a statewide payment locality into multiple localities would generally result in urban physicians experiencing an increase in payment and rural physicians experiencing a decrease in payment. For this reason, we have stated that we would consider a petition from a state medical association that could demonstrate that it had the overwhelming support of physicians in both winning and losings areas. We did not set absolute numerical levels of support because of the uniqueness of the locality structure in each state; we said that setting a numerical level of support would limit the discretion required to properly evaluate each request. We have employed a consistent process in evaluating each request. As stated in the CY 1995 PFS rule (59 FR 63416), upon receiving a preliminary contact from a State medical society, we inform the society that at a minimum we require: (1) A formal request for the change from the State medical society, along with a copy of a recently adopted resolution requesting the change; (2) the number of licensed actively practicing physicians in the State and the number that are society members; (3) the number of society members in each local (county) society; and (4) letters from the local societies representing physicians in the losing areas indicating the level of support for the change. After evaluating this material, if we believe that the material demonstrates overwhelming support among both winning and losing physicians, we announce the proposed change in the Federal Register. If the public comments received demonstrate this overwhelming support, we announce the change in a final rule in the Federal Register. We reiterated in the CY 2010 PFS final rule (74 FR 61758), that in the event we decide to make a specific proposal for changing the locality configuration, we would also provide extensive opportunities for public input (for example, Town Hall meetings or Open Door Forums, as well as opportunities for public comments afforded by the rulemaking process).

Regarding alternative data sources for office rent data, we note that our efforts are ongoing to identify a publicly available, robust, nationally representative commercial rent data source that could be made available to CMS for this purpose. We refer readers to the CY 2023 PFS final rule (87 FR 69625 through 69630) where we undertook a comprehensive analysis of alternatives to the ACS data and concluded that there is still no acceptable national data source available for physician office or other comparable commercial rents, and therefore, we proposed to continue to use county-level residential rent data from the ACS as a proxy for the relative cost differences in commercial office rents for the proposed CY 2023 update, and have done so in calculating the CY 2026 GPCIs as well.

With regard to the supplies, equipment, and miscellaneous expense cost index component of the PE GPCIs, we note that we made no proposals regarding our current policy for this component of the PE GPCI. We have stated that we believe there is a national market for these items and there is not significant geographic variation in those costs, and as such we assign a value of 1.00 for this component for each locality, consistent with the national average. The commenter did not provide any data or information to quantify the variation of costs of supplies, the amount of supplies lost to expiration

dates, or national suppliers' order minimums in contrast to a rural specialty practice's demand for these supplies. We encourage the commenter and other interested parties to submit data supporting their assertions for consideration in future rulemaking; specifically, we would be interested in information regarding potential data sources for shipping costs and the costs of medical equipment and supplies for different geographic regions. Ideally, the potential data sources are accessible to the public, available on a national basis for both urban and rural areas and updated regularly. Similarly, we have previously attempted to locate data sources specific to geographic variation in shipping costs, and we found no comprehensive national data source for this information, and therefore, we have not been able to quantify variation in costs specific to islands or rural communities.

Comment: A few commenters provided feedback on the proposed CY 2026 GPCIs for California. One of the commenters expressed support for the smooth transition to the new physician geographic payment localities in California and supported the proposed GPCIs for 2026 based on accurate data analysis. Another commenter stated concerns that the GPCI fails to capture the true variation in physician compensation and practices in high-cost regions, like Northern California. The commenter also stated that by statute, the work GPCI only accounts for 25 percent of local cost differences compared to the national average, systemically understating costs in high-cost areas.

Response: We appreciate the commenters' feedback and support of our proposals and locality transitions over the last decade. We would also like to reiterate that the GPCIs are based on nationally representative and publicly available wage data from the BLS OEWS for the work GPCI and employee wage and purchased services component of the PE GPCI, and the Census Bureau's ACS data for the rent index component of the PE GPCI. The GPCIs are a measure of the relative resource of cost differences among localities compared to the national average as informed by data (not a measure of absolute costs).

Comment: A commenter stated that the GPCIs for Hawaii do not account for the unique costs of providing medical services in Hawaii, a non-contiguous state. The commenter stated that the work GPCI is flawed for Hawaii due to the State's high cost of living, significant health workforce shortages and issues with the proxy professional wages used by CMS to calculate payments. The commenter also stated that in 2008, section 134 of Public Law 110-275 implemented a 1.5 GPCI work GPCI floor for all counties in the Alaska locality. However, despite facing similar challenges as a non-contiguous state, Hawaii did not receive a similar GPCI adjustment. The commenter recommends that CMS institute a 1.5 work GPCI floor for the Hawaii PFS locality to address these issues and ensure parity for non-contiguous states. The commenter stated that this adjustment would help reflect the true cost of physician services in Hawaii and support the recruitment and retention of healthcare professionals in the state.

Response: We reiterate that the GPCIs are based on nationally represented and publicly-available wage data from the BLS OEWS for the work GPCI. The GPCIs are a measure of relative resource cost differences among localities compared to the national average as informed by the data (not a measure of absolute costs). We remind commenters that the work GPCI value for Alaska is not based on the data for that State, instead section 1848(e)(1)(G) of the Act sets a permanent 1.5 work GPCI floor for Alaska. Similarly, section 1848(e)(1)(I) of the Act sets a permanent PE GPCI floor of 1.0 for the Frontier States. These statutory requirements are State-specific, and therefore we do not have the authority to extend a work GPCI floor to Hawaii.

Comment: A commenter expressed concern about the proposed CY 2026 GPCI decreases for some localities in Texas. The commenter requested that we recalculate the GPCIs for Texas using accurate, localized economic data that reflects the State's economy and inflationary pressures. They also requested that we revise the GPCI methodology to reflect real-world, region-specific economic conditions.

Response: Because Medicare is a national program, and section 1848(e)(1)(A) of the Act requires us to establish GPCIs to measure relative cost differences among localities compared to the national average, we believe it is important to use the best data sources that are available on a nationwide basis. These data sources should be regularly updated and retain consistency area-to-area, year-to-year. We welcome opportunities to discuss alternative data sources with interested parties and to incorporate such data, as appropriate in the GPCI calculation process, through our standard annual rulemaking process.

Comment: A commenter raised concerns about the GPCI for Rhode Island. The commenter stated that Medicare reimburses Rhode Island at the lowest rate compared to Connecticut and both regions of Massachusetts. This has led to the healthcare workforce leaving Rhode Island and increasing financial burdens to practices in Rhode Island. The commenter suggested that CMS review the GPCI for Rhode Island to prevent further loss of professionals, diminishing access to care, and impacts on healthcare costs.

Response: We reviewed the Rhode Island GPCI value and believe it is appropriate based upon the data sources and methodology used to calculate the GPCIs. We would like to reiterate that the GPCIs are based on nationally representative and publicly available wage data from the BLS OEWS for the work GPCI and employee wage and purchased services components of the PE GPCI, and the Census Bureau's ACS data for the rent index component of the PE GPCI. The GPCIs are a measure of relative resource cost differences among localities compared to the national average as informed by the data (not a measure of absolute costs). We welcome opportunities to discuss alternative data sources with interested parties and to incorporate such data, as appropriate in the GPCI calculation process, through our standard annual rulemaking process.

Comment: A commenter stated that the “rural vs non-rural distinctions” in the GPCI locality structure are outdated, especially in the post-pandemic employment landscape. The commenter noted two significant shifts: the predominance of remote work in healthcare and the widespread nature of travel work in clinical areas. Additionally, the commenter stated that Mississippi is the most medically underserved state, with most counties designated as a HPSA. The commenter requested that the current GPCI locality structure be reevaluated and replaced.

Response: We appreciate the commenter's feedback regarding the evolving healthcare landscape and may consider changes to the PFS locality structure through future rulemaking. We welcome any additional feedback from interested parties regarding the changing landscape of healthcare, specifically regarding how to appropriately account for remote work and travel work in the PFS locality structure and GPCI methodology, through our standard annual rulemaking process.

Additionally, section 1848(e)(1)(A) of the Act requires us to develop separate GPCIs to measure resource cost differences among localities compared to the national average for each of the three GPCI components, and section 1848(e)(1)(C) of the Act requires us to

review and, if necessary, adjust the GPCI at least every 3 years. Based on new data, GPCI values may increase or decrease.

Comment: A commenter expressed concerns regarding the disparities in the GPCI that affect healthcare providers in rural and underserved areas. The commenter stated that Iowa ranks 47th out of 50 states for total GPCI, leading to inequitable payment for providers in the region. The commenter recommended that CMS either remove the GPCI altogether or implement a national GPCI floor to ensure minimum equitable payment. The commenter states that this change is essential to support recruitment, retention, and access to care in rural states like Iowa, as the current methodology does not accurately reflect the costs of operating clinical practices in low-GPCI areas and undermines health equity goals.

Response: Section 1848(e)(1)(C) of the Act requires us to review, and if necessary, adjust the GPCIs at least every 3 years. They are a measure of relative resource cost differences among localities compared to the national average as informed by the data. Section 1848(e)(1)(E) of the Act provides for a 1.0 floor for the work GPCIs, which was extended by many successive amendments to the statute through September 30, 2025. Given these statutory requirements, CMS does not have the authority to remove the GPCI altogether or implement a national GPCI floor.

Comment: Several commenters provided feedback on the 2017-based MEI cost share weights and the weights based on Physician Practice Information \123\ (PPI) and Clinician Practice Information \124\ (CPI) Survey data for purposes of alternatives considered for the CY 2026 PE GPCIs and PFS ratesetting and for future rulemaking.

\123\ https://www.ama-assn.org/system/files/table-1-results-from-ppi.pdf.

\124\ https://www.ama-assn.org/system/files/table-1-results-from-cpi-final.pdf.

A couple of commenters suggested using the PPI and CPI data to update the MEI cost share weights. They expressed concerns regarding using outdated 2006-based MEI cost share weights and recommend using more recent data. One of the commenters requested that CMS reconsider using the 2006 MEI cost share weights and incorporate the AMA's Physician Practice Information Survey (PPIS) data into the CY 2026 update. The commenter requested CMS to provide a clear timeline and plan for addressing gaps identified in the PPI and CPI Survey dataset. The other commenter stated for the CY 2026 MEI and GPCI weights, CMS should use PPI and CPI Survey data to implement updated shares of work, practice expense, and professional liability insurance (PLI), which results in the following distribution: work = 54.4 percent; PE = 43.8 percent; and PLI = 1.7 percent. The commenter recommended that the updated CY 2026 GPCIs be phased in over 2 years. Additionally, the commenter recommended that CMS modify its proposed mapping of the PPI and CPI survey data categories for use in updating the PE GPCI.

A couple of commenters support maintaining the current 2006 MEI cost share weights to update the CY 2026 GPCI. One of these commenters reviewed the alternative derived weights from the most recent PPI and CPI survey data for the GPCIs and found them problematic and agreed with CMS' reasons for maintaining the current 2006 MEI cost share weights. The commenter was troubled by the significant difference between the current and 2017-based cost share weights and found that the 50 percent mapping of administrative, overhead and other categories of the PPI and CPI survey to the purchased services and office rent component of the PE GPCI to be arbitrary. The commenter continued to state that the revised cost weights are not supported by other data and cited a recent Medical Group Management Association (MGMA) survey of group practices which estimated provider income (that is, physician work) to account for 30 to 40 percent of revenue, labor (support staff) was the largest component of practice expenses, accounting for 25 to 30 percent of revenue, rent accounting for 5 to 10 percent of costs (6 percent of revenue) and malpractice accounting for 3 to 5 percent of costs. The commenter stated they would not expect CMS to use MGMA data to develop weights but believe proposed weights should align with such findings. Therefore, the 2017-based MEI cost share weights and the weights based on PPI and CPI survey data more closely align with the MGMA data and their understanding of practice costs in California. However, the commenter stated that rather than supporting one of the two proposed MEI alternatives, they recommended that CMS convene a new technical advisory panel to make recommendations for reconciling the differences in work and practice expenses between physician and non- physician owned practices, safeguarding accurate payment to the physician owned practices without overpaying non-physician owned practices. The commenter stated that the technical advisory panel should make recommendations on the practice expense weights and, if there are significant differences in the purchased service component cost shares, the panel should review and make recommendations regarding the labor-related share. In addition, the commenter recommended that updates to the cost share weights occur simultaneously with updates to the GPCIs to provide greater consistency and stability between updates and maintaining the 2-year phase in.

A couple of commenters recommended using the 2017-based MEI cost share weights. A commenter recommended that CMS implement the 2017- based MEI cost share weight with a multi-year transition period to help physician practices and other Part B providers adjust to any redistributive effects. Another commenter expressed concerns with using the 2006-based MEI cost share weights, stating that relying on nearly 20 years old data would significantly mispresent current practice costs. The commenter stated that continuing to base payments on outdated inputs risked widening the gap between actual costs and Medicare payments, threatening the financial sustainability of providers and ultimately access to care. The commenter recommended for CMS to use more recent data sources, such as the 2017-based MEI cost share.

Response: We appreciate the commenters for their support of our proposal and may consider the feedback for possible future rulemaking. We remind commenters that, for the CY 2026 GPCI update, we had concerns about the potential redistributive effects that implementing the 2017- based MEI would have on PFS payments. Additionally, we received data from the AMA's PPI and CPI Surveys. However, these data lack the specific breakdown of practice expense that we needed to consider its use to weight the four components of the PE GPCI for CY 2026. We also note that maintaining the 2006-based MEI cost share weights for the CY 2026 GPCI update preserves consistency in the data used to update both the GPCI and PFS ratesetting inputs for CY 2026.

Comment: Several commenters commended CMS for exploring alternatives to simplify the work GPCI development by reducing the number of occupations used. A commenter stated that of the two scenarios investigated, they favor the 75 percent bachelor's degree or higher and the greater than 75 percent county wage data availability due to its greater simplicity and similar accuracy, which are the criteria we use

to assess alternatives. The commenter questioned why the greater than 90 percent criteria were not reported regarding accuracy relative to the current GPCIs, as these criteria would further simplify the methodology by reducing the number of occupation codes included in the calculations. The commenter stated that if these more stringent criteria had similar accuracy in measuring relative cost differences, the commenter would favor them. Additionally, the commenter requested the specific derived BLS OEWS national level files and cross industry MSA level wage files for the work GPCI, and the wage and purchased services PE GPCI as described in the Interim Report's methodology of Work and PE GPCI formation be made publicly available. The commenter also requested additional information as to whether the purchased services indices included a non-labor related adjustment and, if so, what that adjustment was. The commenter appreciated CMS' willingness to provide additional data to verify the GPCI methodology.

Response: We appreciate the commenters for their support of our efforts and may consider the feedback for possible future rulemaking. We appreciate the commenter's requests for additional information and data files and welcome the opportunity to discuss these requests in more detail with the interested party. We note that we may consider modifications to the list of occupation codes used in the calculation of the work GPCI, which would simplify the calculation and provide opportunity to improve clarity of our documentation.

Comment: A few commenters supported the proposed updates to the GPCIs.

Response: We appreciate the commenters for the support of our proposed CY 2026 GPCI updates.

After consideration of public comments, we are finalizing the CY 2026 GPCI updates as proposed. The final GPCIs and summarized GAFs are displayed in Addenda D and E to this final rule.

1. Requiring Manufacturers of Certain Single-Dose Container or Single- Use Package Drugs To Provide Refunds With Respect to Discarded Amounts (Sec. Sec. 414.902 and 414.940) a. Background

Section 90004 of the Infrastructure Investment and Jobs Act (Pub. L. 117-58, November 15, 2021) (hereinafter referred to as “the Infrastructure Act”) amended section 1847A of the Act to add a provision requiring manufacturers to provide a refund to CMS for certain discarded amounts from a refundable single-dose container or single-use package drug (hereinafter referred to as “refundable drug”) for calendar quarters beginning January 1, 2023.

The calculation of the refund is codified at Sec. 414.940(c). For a new refund quarter (as defined at Sec. 414.902) beginning on or after January 1, 2023, an amount equal to the estimated amount (if any) by which:

The product of the total number of units of the billing and payment code for such drug that were discarded during such new refund quarter; and the amount of payment determined for such drug or biological under section 1847A(b)(1)(B) or (C) of the Act, as applicable, for such new refund quarter;

Exceeds an amount equal to the applicable percentage of the estimated total allowed charges for such drug for the new refund quarter.

Section 1847A(h)(3)(B)(ii) of the Act provides that, in the case of a refundable drug that has unique circumstances involving similar loss of product as that described in section 1847A(h)(8)(B)(ii) of the Act, the Secretary may increase the applicable percentage otherwise applicable as determined appropriate by the Secretary. In the CY 2023 PFS final rule (87 FR 69731), we adopted an increased applicable percentage of 35 percent for drugs reconstituted with a hydrogel and with variable dosing based on patient-specific characteristics. In the CY 2024 PFS final rule (88 FR 79047 through 79064), we finalized an increased applicable percentage for two categories of drugs with unique circumstances, codified at Sec. 414.940(d). These categories include: certain drugs with a low-volume dose (that is, where the volume removed from the vial or container containing the labeled dose does not exceed 0.1 mL or falls between 0.11 mL and 0.4 mL); and orphan drugs furnished to fewer than 100 unique beneficiaries. Drugs with an increased applicable percentage are listed on the CMS website.\125\

\125\ https://www.cms.gov/medicare/payment/part-b-drugs/discarded-drugs.