Announcement of Calendar Year (CY) 2024 Medicare Advantage (MA) Capitation Rates and Part C and Part D Payment Policies

March 31, 2023

NOTE TO: Medicare Advantage Organizations, Prescription Drug Plan Sponsors, and

Other Interested Parties

Announcement of Calendar Year (CY) 2024 Medicare Advantage (MA) Capitation Rates

and Part C and Part D Payment Policies

In accordance with section 1853(b)(1) of the Social Security Act, we are notifying you of the

annual capitation rate for each Medicare Advantage (MA) payment area for CY 2024 and the

risk and other factors to be used in adjusting such rates.

In response to our request for comments on the Advance Notice of Methodological Changes for

CY 2024 MA Capitation Rates and Part C and Part D Payment Policies (CY 2024 Advance

Notice), published on February 1, 2023, CMS received submissions from professional

organizations, MA and Part D sponsors, advocacy groups, physicians, state Medicaid agencies,

pharmaceutical manufacturers, pharmacy benefit managers, pharmacies, and other interested

persons. The Rate Announcement describes and responds to all of the substantive comments

received.

After considering all comments received, we are finalizing policies in the Announcement of CY

2024 MA Capitation Rates and Part C and Part D Payment Policies (CY 2024 Rate

Announcement) that reflect CMS’ commitment to ensuring that people with Medicare receive

equitable, affordable, high quality, and whole-person care now and in the future, especially the

most vulnerable. The policies in the CY 2024 Rate Announcement are an important step in our

efforts to make sure the MA program meets the health care needs of all beneficiaries while

improving the quality and long-term stability of the Medicare program. The CY 2024 Rate

Announcement finalizes an important transition to an updated risk adjustment model that

implements a set of commonsense, clinically-based technical updates needed to keep MA

payments up-to-date and to improve payment accuracy to MA plans.

Specifically, the updated risk adjustment model is developed using ICD-10 codes to align with

the rest of the health care system, which has been using ICD-10 since 2015. It also incorporates

newer data – the current MA risk adjustment model is calibrated with 2014 diagnosis data and

2015 FFS expenditure data and the new model uses 2018 diagnosis data and 2019 expenditure

data. Finally, the revised model includes clinically-based adjustments to ensure that conditions

included in the model are stable predictors of costs. These adjustments help ensure payments

accurately reflect what it costs to care for beneficiaries and make the model less susceptible to

discretionary coding, which can lead to excess payments to MA plans. This is consistent with

updates we have done in the past where we removed or reclassified codes disproportionately

coded in MA compared to Medicare FFS to avoid wasteful spending.

Together, these updates improve the model’s ability to predict the cost of care and ensure MA

risk-adjusted payments are as accurate as possible, which ultimately makes sure MA plans are

paid enough to deliver the benefits that their enrollees are entitled to.

The capitation rate tables for 2024 and supporting data are posted on the CMS website at

https://www.cms.gov/Medicare/Health- Plans/MedicareAdvtgSpecRateStats/Ratebooks-and-

Supporting-Data.html. The statutory component of the regional benchmarks, qualifying counties,

and each county’s applicable percentage are also posted on this section of the CMS website.

Attachment I of the Rate Announcement shows the final estimates of the National Per Capita

MA Growth Percentage for 2024 and the National Medicare Fee-for-Service (FFS) Growth

Percentage for 2024, used to calculate the 2024 capitation rates. As discussed in Attachment I,

the final estimate of the National Per Capita MA Growth Percentage for combined aged and

disabled beneficiaries is 1.60 percent, and the final estimate of the FFS Growth Percentage is

2.45 percent. Attachment II provides a set of tables that summarizes many of the key Medicare

assumptions used in the calculation of the growth percentages.

Section 1853(b)(4) of the Social Security Act requires CMS to release county-specific per capita

FFS expenditure information on an annual basis, beginning with March 1, 2001. In accordance

with this requirement, FFS data for CY 2021 were posted on the above website with the Advance

Notice.

Attachment II details the key assumptions and financial information behind the growth

percentages presented in Attachment I.

Attachment III presents responses to Part C payment-related comments on the CY 2024 Advance

Notice.

Attachment IV presents responses to Part D payment-related comments on the Advance Notice.

Attachment V provides the final Part D benefit parameters and details how they are updated.

Attachment VI presents responses to comments on updates for MA and Part D Star Ratings.

Attachment VII contains economic information for significant provisions in the CY 2024 Rate

Announcement.

Attachment VIII contains the CMS-HCC Risk Adjustment Factor and Predictive Ratio tables.

Key Updates from the Advance Notice

Growth Percentages: Attachment I provides the final estimates of the National Per Capita MA

Growth Percentage and the FFS Growth Percentage, also called growth rates, upon which the

capitation rates are based, and information on deductibles for Medical Savings Accounts. Each

year for the Rate Announcement, CMS updates the growth rates to be based on the most current

estimate of per capita costs, based on the available historical program experience and projected

trend assumptions at that time. The growth rates change between proposed and final as CMS

incorporates updated data and assumptions. This year, the change in growth rates from the

Advance Notice to the Rate Announcement is due to several key factors, including: additional

CY 2022 experience data that was lower than previously projected, updated modeling to account

for the effects of COVID-19 and other programmatic and demographic changes, lower morbidity

from excess COVID-related deaths, lower total spending by explicitly modeling the shift of hip

and knee replacements from inpatient to outpatient setting, and updated modeling of the effect of

a greater share of dual beneficiaries enrolling in MA. For more information on the overall change

in growth rates, please see the Fact Sheet released in the Newsroom section of the CMS.gov

website that accompanies this Rate Announcement.

Technical Update to Medical Education Payments in the Non-ESRD USPCC Baseline:

CMS is finalizing the technical update to remove MA-related indirect medical

education and direct graduate medical education costs (described in the CY 2024

Advance Notice) from the historical and projected expenditures supporting the final

estimates (being released in this Rate Announcement) of the non-ESRD FFS USPCCs.

The Secretary has directed the CMS Office of the Actuary (OACT) to phase in this

technical update to the USPCCs over a 3-year period beginning with the CY 2024

ratebook, with 33% of the medical education adjustment applied to the USPCCs in

2024. We expect 67% of the 2025 medical education adjustment to be applied in 2025

and 100% of the 2026 value to be applied in 2026.

Calculation of FFS Costs: The Secretary has directed the CMS Office of the Actuary to adjust

the FFS experience for beneficiaries enrolled in Puerto Rico to reflect the propensity of “zero–

dollar” beneficiaries nationwide.

CMS-Hierarchical Condition Categories (CMS-HCC) Risk Adjustment Model (Non-PACE):

CMS is finalizing the updated risk adjustment model proposed in the CY 2024 Advance

Notice, but will phase it in over 3 years. For CY 2024, risk scores will be calculated as a blend

of 67% of the risk scores calculated with the current model (the 2020 model) and 33% of the

risk scores calculated with the updated model (the 2024 model). For CY 2025, we expect risk

scores to be calculated as a blend of 33% of the risk scores calculated with the 2020 model and

67% of the risk scores calculated with the 2024 model, and for CY 2026, we expect 100% of

the risk scores to be calculated with the 2024 model.

Frailty Adjustment for Fully Integrated Dual Eligible Special Needs Plans (FIDE SNPs): CMS is

finalizing frailty factors that do not include the Consumer Assessment of Healthcare Providers

and Systems (CAHPS) survey weight, which is an adjustment that can be used to account for

potential non-response bias. This weight was proposed to be included in the frailty factor

calculation in the CY 2024 Advance Notice. CMS will implement FIDE SNP frailty factors

consistent with the updated CMS-HCC risk adjustment model being finalized for CY 2024. Also,

consistent with CMS’ proposal to blend risk scores for CY 2024 (67% current model and 33%

updated model), a blended frailty score for FIDE SNPs will be compared with PACE frailty

calculated in the same manner to determine whether that FIDE SNP has a similar average level

of frailty as PACE. The final frailty factors for CY 2024 can be found in Section L.

Policies Adopted as Described

As in past years, policies in the Advance Notice that are not modified or retracted in the Rate

Announcement become effective for the upcoming payment year. Clarifications in the Rate

Announcement supersede information in the Advance Notice and prior Rate Announcements

as they apply for payment year 2024.

MA Benchmark, Quality Bonus Payments, and Rebate: We will continue to implement the

methodology, as described in the CY 2024 Advance Notice, used to derive the benchmark

county rates, how the qualifying bonus counties are identified, and the applicability of the Star

Ratings.

Location of Network Areas for Private Fee-for-Service (PFFS) Plans in Plan Year 2025: The list

of network areas for plan year 2025 is available on the CMS website at https://www.cms.gov/

Medicare/Health-Plans/PrivateFeeforServicePlans/NetworkRequirements.html.

Direct Graduate Medical Education (DGME) Carve-out Applied to Average Geographic

Adjustments (AGAs): As in past years, we will continue carving out FFS DGME amounts

from the MA capitation rates. (This is different than the technical update related to medical

education payments on behalf of MA enrollees in the Non-ESRD USPCC baseline discussed

above.)

Organ Acquisition Costs for Kidney Transplants: We will continue carving out Kidney

Acquisition Costs (KAC) from the MA capitation rates.

Indirect Medical Education (IME) Phase Out Applied to AGAs: We will continue phasing out

FFS IME amounts from the MA capitation rates.

MA End Stage Renal Disease (ESRD) Rates: We will continue to set MA ESRD rates on a

state basis.

MA Employer Group Waiver Plans (EGWPs): We will continue to use the payment

methodology as described in the Advance Notice, but with the finalized bid-to-benchmark

ratios for 2024 MA EGWP Payment rates as indicated in the table below. These bid-to-

benchmark ratios are weighted by February 2023 enrollment, including retroactive enrollment

adjustments made in March 2023 to the February 2023 enrollment file, due to a system

processing error.

Applicable Percentage

Bid to Benchmark Ratio

0.95

78.5%

77.2%

1.075

76.6%

1.15

76.8%

CMS-HCC Risk Adjustment Model (PACE): For CY 2024, CMS will continue to use the 2017

CMS-HCC risk adjustment model and associated frailty factors to calculate risk scores for

participants in PACE organizations.

CMS-HCC ESRD Risk Adjustment Models:

• For Non-PACE Organizations: For CY 2024, CMS will continue to use the 2023 CMS-

HCC ESRD risk adjustment models to calculate risk scores for beneficiaries in dialysis,

transplant, and post-graft status.

• For PACE Organizations: For CY 2024, CMS will continue to use the 2019 CMS-HCC

ESRD risk adjustment models to calculate risk scores for participants in PACE

organizations with ESRD.

Frailty Adjustment for PACE Organizations: For CY 2024, CMS will continue to use the frailty

factors associated with the 2017 CMS-HCC model (as displayed in Table II-6 of the CY 2024

Advance Notice) to calculate frailty scores for PACE organizations.

Medicare Advantage Coding Pattern Difference Adjustment: For CY 2024, CMS will continue

to apply the statutory minimum MA coding pattern difference adjustment factor of 5.90 percent.

Final 2024 Normalization Factors:

CMS will finalize the 2024 Normalization Factor methodologies as proposed in the Advance

Notice.

For the three CMS-HCC risk adjustment models with a 2019 or 2020 denominator listed below,

CMS will calculate the normalization factors using a five-year linear slope methodology and

updated average FFS risk scores for 2018 through 2022, but continuing to exclude the 2021 risk

score.

• 2024 CMS-HCC model (for non-PACE organizations), for blended risk score

calculations: 1.015

For more information, see the HPMS memorandum dated January 31, 2023 regarding the February 2023 MARx plan payment.

• 2023 CMS-HCC ESRD dialysis model (for non-PACE organizations): 1.022

• 2023 CMS-HCC ESRD functioning graft model (for non-PACE organizations): 1.028

For the four CMS-HCC risk adjustment models with a 2015 denominator listed below and the

RxHCC models, CMS will calculate the normalization factors using a five-year linear slope

methodology and historical FFS risk scores (2016 through 2020), without including the 2021 and

2022 risk scores.

• 2020 CMS-HCC model (for non-PACE organizations), for blended risk score

calculations: 1.146

• 2017 CMS-HCC model (for PACE organizations): 1.159

• 2019 CMS-HCC ESRD dialysis model (for PACE organizations): 1.100

• 2019 CMS-HCC ESRD functioning graft model (for PACE organizations): 1.159

• 2023 RxHCC model (for non-PACE organizations): 1.063

• 2020 RxHCC model (for PACE organizations): 1.084

Sources of Diagnoses for Risk Scores calculated with CMS-HCC and CMS-HCC ESRD Risk

Adjustment Models:

• For Non-PACE organizations: CMS will continue the policy adopted in the CY 2023

Rate Announcement to calculate risk scores for payment to MA organizations and certain

demonstrations using only risk adjustment-eligible diagnoses from encounter data and

FFS claims.

• For PACE organizations: CMS will continue using the same method of calculating risk

scores that we have been using since CY 2015, which is to pool risk adjustment-eligible

diagnoses from the following sources to calculate a single risk score (with no weighting):

(1) encounter data, (2) Risk Adjustment Processing System (RAPS) data, and (3) FFS

claims.

RxHCC Risk Adjustment Models:

• For Non-PACE Organizations: For CY 2024, we will continue to use the 2023 RxHCC

risk adjustment model to adjust direct subsidy payments for Part D benefits offered by

stand-alone Prescription Drug Plans (PDPs) and Medicare Advantage-Prescription

Drug Plans (MA-PDs).

• For PACE Organizations: For CY 2024, CMS will continue to use the 2020 RxHCC

risk adjustment model to calculate Part D risk scores.

Source of Diagnoses for Risk Scores calculated with the RxHCC Risk Adjustment Models:

• For Non-PACE organizations: CMS will continue the policy adopted in the CY 2023

Rate Announcement to calculate Part D risk scores using only risk adjustment-eligible

diagnoses from encounter data and FFS claims.

• For PACE organizations: CMS will continue using the same method of calculating risk

scores that we have been using since CY 2015, which is to pool risk adjustment-eligible

diagnoses from the following sources to calculate a single risk score (with no weighting):

(1) encounter data, (2) RAPS data, and (3) FFS claims.

Inflation Reduction Act Updates: CMS will implement the changes to the Part D drug benefit

made by the Inflation Reduction Act of 2022 as described in the CY 2024 Advance Notice:

• Cost sharing for covered Part D drugs will be eliminated for beneficiaries in the

catastrophic phase of coverage beginning in CY 2024.

• Beginning in CY 2024, the low-income subsidy program (LIS) under Part D will

increase the income limits for the full LIS benefit from 135 percent of the federal poverty

limit (FPL) to 150 percent of the FPL. Medicare beneficiaries earning between 135

percent and 150 percent of the FPL in CY 2024, who meet the resources requirements

under either sections 1860D-14(a)(3)(D) or (E) of the Act, and who would have been

eligible for the partial low-income premium and cost-sharing subsidies and a reduced

deductible under section 1860D-14(a)(2) of the Act had the IRA not been enacted, will be

eligible for full low-income premium and cost-sharing subsidies and a $0 deductible.

• For CY 2024, the deductible will continue not to apply to any Part D covered insulin

product. Also, in the initial coverage phase and the coverage gap phase, cost sharing must

not exceed the applicable copayment amount, which for CY 2024 is $35 for a month’s

supply of each covered insulin product.

• For CY 2024, the deductible will continue not to apply to any adult vaccine

recommended by the Advisory Committee on Immunization Practices (ACIP). Also, the

statute requires these vaccines to be exempt from any co-insurance or other cost sharing,

including cost sharing for vaccine administration and dispensing fees for such products,

when administered in accordance with ACIP’s recommendation, for beneficiaries in the

initial coverage and coverage gap phases.

• Beginning in CY 2024, the base beneficiary premium (BBP) growth will be held to no

more than 6 percent by statute. The BBP for Part D in 2024 will be the lesser of the BBP

The elimination of the deductible for each Part D covered insulin product and implementation of cost-sharing capped at $35 for

a month’s supply of each Part D covered insulin product has been effective as of January 1, 2023. See HPMS Memorandum,

Contract Year 2023 Program Guidance Related to Inflation Reduction Act Changes to Part D Coverage of Vaccines and Insulin,

September 26, 2022.

The elimination of the deductible and cost sharing for any adult vaccine recommended by ACIP has been effective as of

January 1, 2023. See HPMS Memorandum, Contract Year 2023 Program Guidance Related to Inflation Reduction Act Changes

to Part D Coverage of Vaccines and Insulin, September 26, 2022.

for 2023 increased by 6 percent or the amount that would otherwise apply under the

original methodology if the IRA were not enacted.

Annual Adjustments to Medicare Part D Benefit Parameters in 2024: We will update the defined

standard benefit deductible amount, initial coverage limit, and out-of-pocket (OOP) threshold, by

multiplying the CY 2023 amounts by the CY 2024 Annual Percentage Increase (API) and

rounding as specified by the statute.

Part D Calendar Year Employer Group Waiver Plans Prospective Reinsurance Amount: We are

maintaining the Part D Calendar Year EGWP prospective reinsurance policy as discussed in the

CY 2024 Advance Notice. The average per member per month (PMPM) actual reinsurance

amount paid to Part D Calendar Year EGWPs for the most recently reconciled payment year,

which for purposes of CY 2024 is CY 2021, was $71.09.

Part D Risk Sharing: We will apply no changes to the current threshold risk percentages for CY

2024.

Retiree Drug Subsidy Amounts: We will use the same methodology as in prior years to update

the cost threshold and cost limit for qualified retiree prescription drug plans.

/ s /

Meena Seshamani, M.D., Ph.D.

Director, Center for Medicare

I, Jennifer Wuggazer Lazio, am a Member of the American Academy of Actuaries. I meet the

Qualification Standards of the American Academy of Actuaries to render the actuarial opinion

contained in this Rate Announcement. My opinion is limited to the following sections of this

Rate Announcement: The growth percentages and United States per capita cost estimates

provided and discussed in Attachments I, II and III; the qualifying county determination,

calculations of Fee-for-Service cost, direct graduate medical education carve-out, kidney

acquisition cost carve-out, IME phase out, MA benchmarks, EGWP rates, and ESRD rates

discussed in Attachment III; the Medicare Part D Benefit Parameters: Annual Adjustments for

Defined Standard Benefit in 2024 described in Attachments IV and V; and the economic

information contained in Attachment VII. As noted in Attachment III, the Secretary has directed

the CMS Office of the Actuary to phase in the MA medical education technical correction to the

USPCCs that are used in determining the growth percentages.

/ s /

Jennifer Wuggazer Lazio, F.S.A., M.A.A.A.

Director

Parts C & D Actuarial Group

Office of the Actuary

Attachments

2024 RATE ANNOUNCEMENT TABLE OF CONTENTS

Announcement of Calendar Year (CY) 2024 Medicare Advantage (MA) Capitation Rates

and Part C and Part D Payment Policies..............................................................................1

Key Updates from the Advance Notice .....................................................................................2

Policies Adopted as Described...................................................................................................4

Attachment I. Final Estimates of the National Per Capita Growth Percentage and the

National Medicare Fee-for-Service Growth Percentage for Calendar Year 2024.............13

Attachment II. Key Assumptions and Financial Information........................................................15

Attachment III. Responses to Public Comments on Part C Payment Policy.................................28

Section A. General Comments.................................................................................................28

Section B. Estimates of the MA and FFS Growth Percentages for 2024 ................................30

Section C. MA Benchmark, Quality Bonus Payments, and Rebate ........................................44

Section D. Calculation of Fee-for-Service Costs.....................................................................45

Section E. Direct Graduate Medical Education .......................................................................55

Section F. Organ Acquisition Costs for Kidney Transplants...................................................55

Section G. IME Phase Out.......................................................................................................55

Section H. MA ESRD Rates ....................................................................................................55

Section I. MA Employer Group Waiver Plans ........................................................................63

Section J. CMS-HCC Risk Adjustment Model for CY 2024 ..................................................65

Section K. End Stage Renal Disease (ESRD) Risk Adjustment Models for CY 2024 .........115

Section L. Frailty Adjustment for Fully Integrated Dual Eligible Special Needs Plans

(FIDE SNPs) and PACE Organizations .......................................................................116

Section M. Medicare Advantage Coding Pattern Adjustment...............................................120

Section N. Normalization Factors..........................................................................................123

Section O. Sources of Diagnoses for Risk Score Calculation for CY 2024 ..........................127

Attachment IV. Responses to Public Comments on Part D Payment Policy...............................128

Section D. Medicare Part D Benefit Parameters: Annual Adjustments for Defined

Section E. Part D Calendar Year Employer Group Waiver Plans Prospective

Section A. RxHCC Risk Adjustment Model .........................................................................128

Section B. Sources of Diagnoses for Part D Risk Score Calculation for CY 2024 ...............130

Section C. Inflation Reduction Act of 2022 Part D Benefit Design Changes .......................130

Standard Benefit in 2024..............................................................................................131

Reinsurance Amount ....................................................................................................132

Section F. Part D Risk Sharing ..............................................................................................133

Attachment V. Final Updated Part D Benefit Parameters for Defined Standard Benefit, Low-

Income Subsidy, and Retiree Drug Subsidy ....................................................................134

Section A. Annual Percentage Increase in Consumer Price index ........................................137

Section B. Calculation Methodology.....................................................................................137

Section C. Annual Percentage Increase in Average Expenditures for Part D Drugs per

Eligible Beneficiary (API) ...........................................................................................139

Section D. Estimated Total Covered Part D Spending at Out-of-Pocket Threshold for

Applicable Beneficiaries ..............................................................................................140

Section E. Retiree Drug Subsidy Amounts............................................................................143

Attachment VI. Updates for Part C and D Star Ratings ..............................................................144

Part C and D Star Ratings and Future Measurement Concepts .............................................144

Reminders for 2024 Star Ratings...........................................................................................144

Measure Updates for 2024 Star Ratings ................................................................................145

Extreme and Uncontrollable Circumstances Policy for the 2024 Star Ratings .....................150

Changes to Existing Star Ratings Measures for the 2023 Measurement Year and Beyond..152

Display Measures...................................................................................................................162

Potential New Measure Concepts and Methodological Enhancements for Future Years .....169

Attachment VII. Economic Information for the CY 2024 Rate Announcement .........................179

Section A. Changes in the Payment Methodology for Medicare Advantage and PACE

for CY 2024..................................................................................................................179

A1. Medicare Advantage and PACE non-ESRD Ratebook ............................................179

A2. Indirect Medical Education (IME) Phase Out...........................................................179

A3. Medicare Advantage and PACE ESRD Ratebooks ..................................................180

A4. CMS-HCC Risk Adjustment Model .........................................................................180

A5. ESRD Risk Adjustment ............................................................................................181

A6. Frailty Adjustment for FIDE SNPs...........................................................................181

A7. MA Coding Pattern Adjustment ...............................................................................181

A8. Normalization............................................................................................................181

Section B. Changes in the Payment Methodology for Medicare Part D for CY 2024 ..........181

B1. Part D Risk Adjustment Model .................................................................................181

B2. Annual Percentage Increase for Part D Parameters ..................................................182

Attachment VIII. CMS-HCC Risk Adjustment Factors & Predictive Ratio Tables....................183

Table VIII-2. 2024 CMS-HCC Model Relative Factors for Aged and Disabled New

Table VIII-3. 2024 CMS-HCC Model Relative Factors for New Enrollees in Chronic

Table VIII-5. Predictive Ratios by Deciles of Predicted Risk (sorted low to high): Non-

Table VIII-6. Predictive Ratios by Deciles of Predicted Risk (sorted low to high): Non-

Table VIII-1. 2024 CMS-HCC Model Relative Factors for Continuing Enrollees...............183

Enrollees.......................................................................................................................193

Condition Special Needs Plans (C-SNPs)....................................................................194

Table VIII-4. 2024 CMS-HCC Model with Disease Hierarchies..........................................195

Dual, Aged (Age >=65) Continuing Enrollee ..............................................................197

Dual, Disabled (Age <65) Continuing Enrollee...........................................................197

Table VIII-7. Predictive Ratios by Deciles of Predicted Risk (sorted low to high): Full

Benefit Dual, Aged (Age >=65) Continuing Enrollee .................................................198

Table VIII-8. Predictive Ratios by Deciles of Predicted Risk (sorted low to high): Full

Benefit Dual, Disabled (Age <65) Continuing Enrollee ..............................................198

Table VIII-9. Predictive Ratios by Deciles of Predicted Risk (sorted low to high): Partial

Benefit Dual, Aged (Age >=65) Continuing Enrollee .................................................199

Table VIII-10. Predictive Ratios by Deciles of Predicted Risk (sorted low to high):

Partial Benefit Dual, Disabled (Age <65) Continuing Enrollee...................................199

Table VIII-11. Predictive Ratios by Deciles of Predicted Risk (sorted low to high):

Institutional Continuing Enrollee .................................................................................200

Aged + Disabled

Dialysis–only ESRD

Current projected 2024 FFS USPCC

$1,105.10

$9,544.97

Prior projected 2023 FFS USPCC

1,078.63

9,332.69

Percent change

2.45 %

2.27 %

Attachment I. Final Estimates of the National Per Capita Growth Percentage and the

National Medicare Fee-for-Service Growth Percentage for Calendar Year 2024

Table I-1 below shows the National Per Capita MA Growth Percentage (NPCMAGP) for 2024.

An adjustment of -1.77 percent for the combined aged and disabled cohort is included in the

NPCMAGP to account for corrections to prior years’ estimates as required by section

1853(c)(6)(C). The combined aged and disabled change is used in the development of the

ratebook.

Table I-1. Increase in the National Per Capita MA Growth Percentages (NPCMAGP) for

2024

Prior increases

Current increases

NPCMAGP for 2024

with §1853(c)(6)(C)

adjustment

2003 to 2023

2023 to 2024

2003 to 2024

Aged + Disabled

109.238 %

105.537 %

3.432 %

112.590%

1.60 %

Current increases for 2003-2024 divided by the prior increases for 2003-2023.

Table I-2 below provides the change in the FFS United States Per Capita Cost (USPCC), which

was used in the development of the county benchmarks. The percentage change in the FFS

USPCC is shown as the current projected FFS USPCC for 2024 divided by projected FFS

USPCC for 2023 as estimated in the 2023 Rate Announcement released on April 04, 2022.

Table I-2. FFS USPCC Growth Percentage for CY 2024

Table I-3 below shows the monthly actuarial value of the Medicare deductible and coinsurance

for 2023 and 2024. In addition, for 2024, the actuarial value of deductibles and coinsurance is

being shown for non-ESRD only, since MA plan bids for 2024 exclude costs for ESRD

enrollees. These data were furnished by the Office of the Actuary.

Table I-3. Monthly Actuarial Value of Medicare Deductible and Coinsurance for 2023 and

2024

2023

2024

Change

2024 non-ESRD

Part A Benefits

$38.18

$36.62

-4.1%

$35.36

Part B Benefits

154.95

161.71

4.4

154.36

Total Medicare

193.13

198.33

2.7

189.72

Includes the amounts for outpatient psychiatric charges.

Medical Savings Account (MSA) Plans. The maximum deductible for MSA plans for 2024 is

$16,000.

Attachment II. Key Assumptions and Financial Information

The USPCCs are the basis for the National Per Capita MA Growth Percentage. Below is a table

that compares last year’s estimates of USPCCs with current estimates for 2003 to 2025. In

addition, this table shows the current projections of the USPCCs through 2026. We are also

providing a set of tables that summarize many of the key Medicare assumptions used in the

calculation of the USPCCs. Most of the tables include information for the years 2003 through

2026.

Most of the tables in this attachment present combined aged and disabled non-ESRD data. The

ESRD information presented is for the combined aged-ESRD, disabled-ESRD, and ESRD only.

All of the information provided in this attachment applies to the Medicare Part A and Part B

programs. Caution should be employed in the use of this information. It is based upon

nationwide averages, and local conditions can differ substantially from conditions nationwide.

None of the data presented here pertain to the Medicare Part D prescription drug benefit.

Table II-1. Comparison of Current & Previous Estimates of the Total USPCC – Non-ESRD

Part A

Part B

Part A + Part B

Calendar

year

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Ratio

2003

$296.18

$247.66

$543.84

1.000

2004

314.08

271.06

585.14

1.000

2005

334.83

292.86

627.69

1.000

2006

345.30

313.70

659.00

1.000

2007

355.44

330.68

686.12

1.000

2008

371.90

351.04

722.94

1.000

2009

383.91

367.49

367.35

751.40

751.26

1.000

2010

383.93

376.34

376.12

760.27

760.05

1.000

2011

387.73

385.30

385.12

773.03

772.85

1.000

2012

377.37

391.93

391.76

769.30

769.13

1.000

2013

380.03

398.72

398.54

778.75

778.57

1.000

2014

370.23

418.36

418.18

788.59

788.41

1.000

2015

373.86

373.99

435.00

434.95

808.86

808.94

1.000

2016

377.62

377.61

444.28

444.14

821.90

821.75

1.000

2017

383.09

382.91

459.19

459.08

842.28

841.99

1.000

2018

388.12

388.06

489.65

489.43

877.77

877.49

1.000

2019

400.79

400.21

521.89

521.77

922.68

921.98

1.001

2020

403.90

402.19

522.48

522.62

926.38

924.81

1.002

Part A

Part B

Part A + Part B

Calendar

year

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Ratio

2021

409.38

412.79

569.14

573.53

978.52

986.32

0.992

2022

431.47

447.39

603.83

624.52

1,035.30

1,071.91

0.966

2023

459.23

469.56

658.56

668.36

1,117.79

1,137.92

0.982

2024

464.05

488.33

692.10

707.07

1,156.15

1,195.40

0.967

2025

480.98

509.50

729.01

744.57

1,209.99

1,254.07

0.965

2026

496.85

772.41

1,269.26

Table II-2. Comparison of Current &

Estimates of the FFS USPCC – Non-ESRD

Part A

Part B

Part A + Part B

Calendar

year

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Ratio

2010

$369.60

$371.20

$374.30

$373.99

$743.90

$745.19

0.998

2011

369.45

371.15

383.17

382.92

752.62

754.07

0.998

2012

355.15

356.97

390.70

390.45

745.85

747.42

0.998

2013

361.78

363.75

394.49

394.24

756.27

757.99

0.998

2014

362.07

364.24

409.16

408.89

771.23

773.13

0.998

2015

366.98

369.37

428.06

427.73

795.04

797.10

0.997

2016

369.00

371.57

433.62

433.36

802.62

804.93

0.997

2017

370.97

373.64

448.28

448.06

819.25

821.70

0.997

2018

374.54

377.84

474.15

473.79

848.69

851.63

0.997

2019

380.01

383.05

500.82

500.77

880.83

883.82

0.997

2020

370.93

372.68

473.65

473.99

844.58

846.67

0.998

2021

384.05

388.34

550.73

546.76

934.78

935.10

1.000

2022

398.10

424.46

573.64

598.85

971.74

1,023.31

0.950

2023

428.63

448.03

629.07

630.60

1,057.70

1,078.63

0.981

2024

440.70

465.39

664.40

666.68

1,105.10

1,132.07

0.976

2025

451.09

484.86

698.89

701.28

1,149.98

1,186.14

0.970

2026

459.88

739.42

1,199.30

Table II-3. Comparison of Current & Previous Estimates of the ESRD Dialysis-only FFS

USPCC

Part A

Part B

Part A + Part B

Calendar

year

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Ratio

2010

$2,952.75

$3,881.39

$6,834.14

1.000

Part A

Part B

Part A + Part B

Calendar

year

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Current

estimate

Last year’s

estimate

Ratio

2011

2,862.38

3,908.01

6,770.39

1.000

2012

2,774.49

3,944.59

6,719.08

1.000

2013

2,794.19

4,088.66

6,882.85

1.000

2014

2,784.52

4,115.70

6,900.22

1.000

2015

2,775.84

4,060.87

6,836.71

1.000

2016

2,895.91

4,081.27

6,977.18

1.000

2017

2,883.27

4,102.66

6,985.93

1.000

2018

2,952.21

4,526.09

7,478.30

1.000

2019

3,040.74

4,614.18

7,654.92

1.000

2020

3,082.55

4,542.51

7,625.06

1.000

2021

3,295.54

3,264.12

4,786.27

5,025.52

8,081.81

8,289.64

0.975

2022

3,395.47

3,646.65

4,863.56

5,279.76

8,259.03

8,926.41

0.925

2023

3,632.99

3,890.68

5,296.62

5,442.01

8,929.61

9,332.69

0.957

2024

3,835.56

4,057.82

5,709.41

5,648.71

9,544.97

9,706.53

0.983

2025

4,084.94

4,242.66

6,778.51

6,426.56

10,863.45

10,669.22

1.018

2026

4,347.69

7,309.00

11,656.69

Table II-4. Basis for ESRD Dialysis-only FFS USPCC Trend

Part A

Part B

Part A & Part B

Calendar

year

All ESRD

cumulative

FFS trend

Adjustment

factor for

dialysis-

only

Adjusted

dialysis-only

cumulative

trend

All ESRD

cumulative

FFS trend

Adjustment

factor for

dialysis-only

Adjusted

dialysis-only

cumulative

trend

All ESRD

cumulative

FFS trend

Adjustment

factor for

dialysis-only

Adjusted

dialysis-only

cumulative

trend

2022 1.02556 1.00465 1.03032 1.01381 1.00230 1.01615 1.01860 1.00327 1.02193

2023 1.09223 1.00931 1.10240 1.10155 1.00461 1.10663 1.09775 1.00652 1.10490

2024 1.14780 1.01400 1.16386 1.18467 1.00693 1.19287 1.16963 1.00976 1.18104

2025 1.21677 1.01871 1.23954 1.40326 1.00925 1.41624 1.32722 1.01278 1.34419

2026 1.28905 1.02344 1.31926 1.50961 1.01157 1.52708 1.41967 1.01597 1.44234

Table II-5. Summary of Key Projections

Part A

Year

Calendar year

CPI percent change

FY inpatient

PPS update factor

FY Part A total reimbursement

(incurred)

2003 1.4% 3.0%

3.5%

2004 2.1 3.4

8.4

2005 2.7 3.3

8.8

2006 4.1 3.7

5.9

2007 3.3 3.4

5.7

2008 2.3 2.7

7.6

2009 5.8 2.7

6.7

2010 0.0 1.9

3.0

2011 0.0 −0.6

4.5

2012 3.6 −0.1

0.4

2013 1.7 2.8

4.7

2014 1.5 0.9

0.6

2015 1.7 1.4

3.2

2016 0.0 0.9

4.3

2017 0.3 0.2

4.0

2018 2.0 1.8

4.0

2019 2.8 1.9

5.5

2020 1.6 3.1

3.2

2021 1.3

2.9

4.8

2022 5.9

2.5

4.7

2023 8.7

4.3

8.1

2024 3.3

2.8

5.3

2025 1.4

3.1

6.8

2026 2.1 2.9

7.3

Part B

Physician fee schedule

Calendar year Fees

Residual

Outpatient hospital

ESRD dialysis update

factor

Total

2003

1.4%

4.5%

4.4%

6.8%

2004

3.8

5.9

11.1

9.8

2005

2.1

3.2

10.8

7.0

2006

0.2

4.6

5.1

6.1

2007

-1.4

3.5

8.2

4.3

2008

-0.3

4.0

6.3

4.8

2009

1.4

2.3

5.4

3.9

2010

2.3

2.1

6.6

2.4

2011

0.8

2.3

7.1

2.5%

2.3

2012

-1.2

0.8

7.2

2.1

1.7

2013

-0.1

0.2

7.2

2.3

0.8

2014

0.4

0.6

12.6

2.8

3.4

2015

-0.3

7.4

0.0

2.7

2016

-0.4

-0.3

5.2

0.15

1.9

2017

0.1

1.1

7.4

0.55

2.8

2018

0.5

1.1

8.4

0.3

5.7

2019

1.2

2.8

4.9

1.3

5.8

2020

0.2

-11.5

-6.0

1.7

-1.3

2021

4.8

13.1

20.0

1.6

8.7

2022

-1.1

4.0

6.5

1.9

4.9

2023

-0.5

3.4

15.1

3.0

7.4

2024

-1.7

3.5

9.1

1.6

4.8

2025

-1.9

3.0

8.3

2.3

5.6

2026

0.4

2.6

8.3

2.1

5.8

Percent change over prior year.

Percent change in charges per aged Part B enrollee.

Reflects the physician update and legislation affecting physician services—for example, the addition of new preventive services enacted in

1997, 2000, and 2010.

Residual factors are factors other than price, including volume of services, intensity of services, and age/sex changes.

The ESRD Prospective Payment System was implemented in 2011.

Part A

Part B

Calendar year

Aged

Disabled

Aged

Disabled

2003

34.437

5.961

33.038

5.215

2004

34.849

6.283

33.294

5.486

2005

35.257

6.610

33.621

5.776

2006

35.795

6.889

33.975

6.017

2007

36.447

7.167

34.465

6.245

2008

37.378

7.362

35.140

6.438

2009

38.257

7.574

35.832

6.664

2010

39.091

7.832

36.516

6.938

2011

39.950

8.171

37.247

7.254

2012

41.687

8.411

38.546

7.502

2013

43.087

8.629

39.779

7.732

2014

44.533

8.776

41.063

7.894

2015

45.911

8.853

42.311

7.977

2016

47.370

8.862

43.623

7.990

2017

48.893

8.940

44.944

8.007

2018

50.457

8.696

46.310

7.862

2019

52.110

8.531

47.765

7.735

2020

53.684

8.319

49.225

7.573

2021

55.041

8.054

50.513

7.359

2022

56.468

7.703

51.923

7.072

2023

58.030

7.384

53.481

6.785

2024

59.651

7.160

55.023

6.586

2025

61.276

7.093

56.595

6.532

2026

62.914

7.098

58.163

6.537

Non-ESRD Total

Table II-6. Medicare Enrollment Projections (In millions)

Non-ESRD Fee-for-Service

Part A

Part B

Calendar year Aged Disabled Aged Disabled

2003

29.593

5.628

28.097

4.875

2004

29.946 5.931 28.300 5.128

2005

30.014 6.178 28.287 5.339

2006

29.362 6.149 27.459 5.270

2007

28.838 6.225 26.782 5.297

2008

28.613 6.241 26.301 5.311

2009

28.563 6.288 26.071 5.374

2010

28.903 6.455 26.261 5.556

2011

29.210 6.659 26.440 5.736

2012

29.960 6.693 26.744 5.779

2013

30.330 6.691 26.948 5.790

2014

30.603 6.618 27.060 5.732

2015

30.947 6.488 27.274 5.609

2016

31.629 6.378 27.814 5.503

2017

31.916 6.299 27.882 5.361

2018

32.168 5.867 27.926 5.028

2019

32.456 5.467 28.017 4.666

2020

32.221 4.953 27.666 4.202

2021

31.438 4.409 26.816 3.711

Part A

Part B

Calendar year

Aged

Disabled

Aged

Disabled

2022

30.802 3.856 26.163 3.224

2023

30.410 3.205 25.703 2.581

2024

30.504 2.652 25.739 2.066

2025

30.941 2.434 26.129 1.864

2026

31.365 2.252 26.482 1.681

ESRD

ESRD - Total

ESRD - Fee-for-Service

Calendar year

Total Part A

Total Part B

Total Part A

Total Part B

2003

0.340

0.331

0.319

0.309

2004

0.353

0.342

0.332

0.321

2005

0.366

0.355

0.344

0.332

2006

0.382

0.370

0.353

0.340

2007

0.396

0.383

0.361

0.347

2008

0.411

0.397

0.367

0.353

2009

0.426

0.412

0.374

0.360

2010

0.442

0.428

0.388

0.373

2011

0.429

0.416

0.371

0.358

2012

0.441

0.429

0.379

0.366

2013

0.454

0.441

0.385

0.372

2014

0.469

0.456

0.390

0.377

2015

0.482

0.468

0.393

0.379

2016

0.496

0.481

0.400

0.384

2017

0.511

0.495

0.404

0.386

2018

0.524

0.507

0.405

0.387

2019

0.537

0.519

0.406

0.387

2020

0.540

0.522

0.396

0.377

2021

0.527

0.511

0.325

0.308

2022

0.509

0.495

0.272

0.258

2023

0.510

0.499

0.245

0.232

2024

0.520

0.508

0.236

0.223

2025

0.531

0.519

0.236

0.223

2026

0.543

0.531

0.237

0.223

Table II-7. Part A Projections for non-ESRD (Aged+Disabled)*

Calendar

year

Inpatient

hospital

Home health

agency SNF Managed care

Hospice: Total

imbursement

(in millions)

2003

2,594.78

370.63

124.28

457.87

5,733

2004

2,714.57

413.44

133.89

500.73

6,832

2005

2,818.21

450.54

140.87

602.29

8,016

2006

2,755.32

475.07

141.30

766.75

9,368

2007

2,696.33

504.24

143.72

916.90

10,518

2008

2,682.50

536.68

151.00

1,088.37

11,404

2009

2,637.34

551.67

153.86

1,260.14

12,274

2010

2,612.51

571.74

155.18

1,264.21

13,126

2011

2,570.82

623.31

138.31

1,314.41

13,897

2012

2,473.46

541.69

130.82

1,376.07

15,068

2013

2,468.49

540.47

128.47

1,416.56

15,263

2014

2,406.24

534.37

123.89

1,372.06

15,346

2015

2,388.54

530.99

126.08

1,435.16

16,159

2016

2,405.32

504.84

121.44

1,495.94

17,128

2017

2,381.77

484.69

117.35

1,609.39

18,228

2018

2,352.67

465.63

113.87

1,721.35

19,570

2019

2,314.56

444.32

108.47

1,938.76

21,174

2020

2,137.02

450.75

95.42

2,160.64

22,319

2021

2,117.54

421.03

93.28

2,277.75

23,034

2022

2,066.30

440.14

88.44

2,579.80

24,195

2023

2,146.67

417.69

94.63

2,848.53

26,017

2024

2,141.39

403.73

95.65

2,924.58

27,867

2025

2,122.61

434.96

102.09

3,108.74

30,079

2026

2,110.05

449.31

108.55

3,290.82

32,605

*Average reimbursement per enrollee on an incurred basis.

Table II-8. Part B Projections for non-ESRD (Aged+Disabled)*

Calendar year

Physician fee schedule Outpatient hospital Durable medical equipment

2003

$1,226.51

$364.77 $196.96

2004

1,344.01

418.85

195.61

2005

1,397.43

477.65

196.83

2006

1,396.40

497.47

197.78

2007

1,368.35

526.92

195.68

2008

1,367.83

555.09

200.92

2009

1,386.03

587.61

183.61

2010

1,429.74

623.13

183.76

2011

1,459.64

662.97

175.84

2012

1,412.72

697.86

173.70

2013

1,369.64

735.35

152.53

2014

1,351.32

823.34

128.57

2015

1,336.26

876.01

132.77

2016

1,313.75

911.03

120.73

2017

1,293.54

952.81

112.30

2018

1,285.36

1,000.20

127.05

2019

1,300.14

1,018.61

129.06

2020

1,111.11

914.01

123.59

2021

1,251.85

1,031.58

121.33

2022

1,208.99

1,014.01

125.84

2023

1,171.90

1,134.23

131.99

2024

1,149.61

1,190.45

133.58

2025

1,139.05

1,266.88

138.33

2026

1,147.58

1,346.13

143.96

Calendar

year

Carrier lab

Physician

administered drugs

Other carrier

Intermediary lab

2003

$73.73

$182.58

$147.21

$75.18

2004

78.48

195.20

158.78

2005

80.47

82.71

178.77

184.02

84.16

2006

85.59

185.41

175.66

84.51

2007

90.65

186.97

176.55

84.38

2008

94.50

184.43

182.19

85.78

2009

101.60

196.19

178.46

79.19

2010

103.81

196.41

178.67

80.23

2011

103.85

209.50

179.44

83.31

2012

111.73

209.34

185.17

84.64

2013

111.79

216.91

177.08

81.74

2014

117.60

224.56

173.55

55.45

2015

113.99

252.11

174.94

55.26

2016

100.91

271.45

172.90

56.21

2017

100.65

280.51

177.43

54.99

2018

107.28

304.36

176.15

52.94

2019

108.73

329.41

174.14

50.31

2020

109.18

325.03

166.87

51.77

2021

122.82

340.52

165.36

56.25

2022

112.53

359.41

178.83

52.95

2023

116.31

368.95

184.37

52.89

2024

118.37

389.08

184.22

52.28

2025

128.70

410.19

187.50

54.52

2026

132.58

432.42

191.43

55.03

*Average reimbursement per enrollee on an incurred basis.

Calendar year

Other intermediary

Home health agency

Managed care

2003

$113.99

$136.75

$421.40

2004

119.58

156.45

471.37

2005

139.78

179.44

560.31

2006

142.09

202.88

769.94

2007

151.16

232.33

931.18

2008

158.20

252.43

1,104.26

2009

187.44

282.09

1,203.79

2010

193.08

283.25

1,221.29

2011

198.15

254.42

1,276.29

2012

205.08

239.36

1,368.13

2013

194.43

234.07

1,497.49

2014

200.51

227.73

1,703.30

2015

210.37

224.84

1,829.20

2016

214.14

219.09

1,938.69

2017

220.58

208.93

2,096.95

2018

228.23

206.47

2,376.35

2019

236.12

201.48

2,704.42

2020

208.83

187.27

3,062.48

2021

219.92

182.59

3,327.89

2022

214.19

169.26

3,800.44

2023

212.86

177.44

4,341.49

2024

214.10

175.38

4,687.26

2025

221.29

182.19

5,007.97

2026

228.98

194.55

5,384.17

* Average reimbursement per enrollee on an incurred basis.

Table II-9. 2024 Projections by Service Category for non-ESRD (Aged+Disabled)*

Service type

Current

estimate

Last year’s

estimate Ratio

Part A

Inpatient hospital

$2,141.39

$2,303.60

0.930

SNF

403.73

447.80

0.902

Home health agency

95.65

126.49

0.756

Managed care

2,924.58

2,978.53

0.982

Part B

Physician fee schedule

1,149.61

1,170.57

0.982

Outpatient hospital

1,190.45

1,232.65

0.966

Durable medical equipment

133.58

126.40

1.057

Carrier lab

118.37

107.45

1.102

Physician Administered Drugs

389.08

420.58

0.925

Other carrier

184.22

163.46

1.127

Intermediary lab

52.28

44.36

1.179

Other intermediary

214.10

245.56

0.872

Home health agency

175.38

245.69

0.714

Managed care

4,687.26

4,716.27

0.994

* Average reimbursement per enrollee on an incurred basis.

Table II-10. Claims Processing Costs as a Fraction of Benefits

Calendar

year

FFS Part A

FFS Part B

Total Part A

Total Part B

2003

0.001849

0.011194

0.001849

0.011194

2004

0.001676

0.010542

0.001676

0.010542

2005

0.001515

0.009540

0.001515

0.009540

2006

0.001245

0.007126

0.001245

0.007126

2007

0.000968

0.006067

0.000968

0.006067

2008

0.000944

0.006414

0.000944

0.006414

2009

0.000844

0.005455

0.000844

0.005455

2010

0.000773

0.005055

0.000773

0.005055

2011

0.000749

0.004396

0.000749

0.004396

2012

0.001008

0.003288

0.001008

0.003288

2013

0.000994

0.002846

0.000994

0.002846

2014

0.001003

0.002884

0.001003

0.002884

2015

0.000952

0.002730

0.000952

0.002730

2016

0.000852

0.002348

0.000852

0.002348

2017

0.000833

0.002111

0.000833

0.002111

2018

0.000836

0.001953

0.000836

0.001953

2019

0.000699

0.001644

0.000699

0.001644

2020

0.000625

0.001536

0.000625

0.001536

2021

0.001038

0.002708

0.000600

0.001399

2022

0.001094

0.002801

0.000582

0.001310

2023

0.001094

0.002801

0.000582

0.001310

2024

0.001094

0.002801

0.000582

0.001310

2025

0.001094

0.002801

0.000582

0.001310

2026

0.001094

0.002801

0.000582

0.001310

Approximate Calculation of the USPCC, the National MA Growth Percentage for

Combined (Aged+Disabled) Beneficiaries, and the FFS USPCC (Aged+Disabled)

The following procedure will approximate the actual calculation of the USPCCs from the

underlying assumptions for the contract year for both Part A and Part B.

Part A: The Part A USPCC can be approximated by using the assumptions in the tables titled

“Part A Projections for non-ESRD (Aged+Disabled)” and “Claims Processing Costs as a

Fraction of Benefits.” Information in the “Part A Projections” table is presented on a calendar

year per capita basis. First, add the per capita amounts over all types of providers (excluding

hospice). Next, multiply this amount by 1 plus the loading factor for administrative expenses

from the “Claims Processing Costs” table. Then, divide by 12 to put this amount on a monthly

basis.

Part B: The Part B USPCC can be approximated by using the assumptions in the tables titled

“Part B Projections for non-ESRD (Aged+Disabled)” and “Claims Processing Costs as a

Fraction of Benefits.” Information in the “Part B Projections” table is presented on a calendar

year per capita basis. First, add the per capita amounts over all types of providers. Next, multiply

by 1 plus the loading factor for administrative expenses from the “Claims Processing Costs”

table and then divide by 12 to put this amount on a monthly basis.

The National Per Capita MA Growth Percentage: The National Per Capita MA Growth

Percentage for 2024 (before adjusting for prior years’ over/under estimates) is calculated by

adding the USPCCs for Part A and Part B for 2024 and then dividing by the sum of the current

estimates of the USPCCs for Part A and Part B for 2023.

The FFS USPCC: The tables used to calculate the total USPCC can also be used to approximate

the calculation of the FFS USPCC. The per capita data presented by type of provider in the

projections tables for both Part A and Part B are based on total enrollment. To approximate the

FFS USPCCs, first add the corresponding provider types under Part A and Part B separately. For

the FFS calculations, do not include the managed care provider type. Next, rebase the sum of the

per capita amounts for FFS enrollees, i.e., multiply the sum by total enrollees and divide by FFS

enrollees. (The enrollment tables in this attachment now also include FFS enrollment.) Then,

multiply by 1 plus the loading factor for administrative expenses and divide by 12. The result

will only be approximate because there is an additional adjustment to the FFS data which

accounts for cost plan data which comes through the FFS data system. This cost plan data is in

the total per capita amounts by type of provider, but it is removed for the FFS calculations.

Attachment III. Responses to Public Comments on Part C Payment Policy

Section A. General Comments

Comment: CMS received a large number of comments in response to the CY 2024 Advance

Notice, with many supporting the direction of the proposals in the Advance Notice and others

expressing concerns about the impacts of the proposed updates. Commenters who supported the

proposals in the Advance Notice believed that the methods taken to update factors, including the

capitation rates, were sound and that the adjustments, such as those to capitation rates, were

much needed to ensure payment accuracy and preserve the financial integrity of the Medicare

Trust Fund. Commenters also cited overpayments to MA organizations and concerns that

wasteful spending is weakening the Medicare program, placing extra pressure on the Medicare

Trust Fund, and being subsidized by all Medicare beneficiaries via higher Part B premiums. A

number of commenters cited MedPAC’s findings that excess payments to MA organizations will

exceed $27 billion in 2023 and they estimate aggregate excess payments to MA organizations

have totaled nearly $124 billion from 2007 to 2023. One commenter stated that this year’s

projected revenue increase of 1% was an improvement in CMS’ management of the MA

program compared to recent years, such as the projected increase of 8.5% for payment in 2023.

Commenters also emphasized that discretionary coding and perceived ‘gaming’ of codes is a key

part of driving excess payments, and applauded efforts to make the model more accurate and less

susceptible to discretionary coding. Another commenter noted that, while changes to the risk

adjustment model in the proposal and previous policies impacting quality bonuses would tend to

reduce payments to MA organizations, these decreases would be more than offset by an increase

in MA benchmarks driven by increases in diagnostic coding by plans. It was also stated that

though many MA organizations and associations state that if the MA risk score trend is excluded

from the year over year revenue calculation projected payment would decrease, there is no

reason to exclude the risk score growth from the estimate of the year-over-year percentage

change in payment to have an accurate estimate.

The commenters who did not support the proposed changes saw the net effect of the proposals as

cuts to the MA program and urged CMS to maintain stability in the program and not implement

the proposed updates. Many commenters submitted a wide array of suggestions regarding

proposals in the Advance Notice, including the medical education adjustment to the growth

percentages and the updated risk adjustment model. Several commenters stated that the

combined effect of the updates to the growth percentages and the risk adjustment model, as well

as the impact of Star Ratings, would materially affect benchmarks and result in significant

payment reductions. Some commenters stated that the MA risk score trend is not actually part of

the MA payment methodology, and that if the risk score trend were not taken into account the

year-to-year change in payment would be a negative 2.27%, and not a positive 1.03%. These

commenters asserted that a variety of impacts would occur if CMS finalized the proposed

changes including that these proposals would disrupt care, place the health of sicker, lower-

income enrollees at risk, and result in increased costs and reduced benefits for MA enrollees.

These commenters highlighted the potential for reduced funding for supplemental benefits and

claimed that these policies would result in reductions in the coverage of dental, hearing, vision,

transportation, and cost sharing for drugs, as well as in hindering the development of complex,

innovative solutions to providing care to Medicare beneficiaries. Further, some commenters

stated that provision of care would suffer and value-based models would also suffer and result in

deteriorating health outcomes.

Response: CMS thanks commenters for their thoughts and input regarding payments made under

the MA program. CMS has a duty to be a steward of the Medicare program. Protecting and

strengthening Medicare for the 65 million Americans who have it now, and all the beneficiaries

in the future, is a key priority for CMS.

Core to this mission is to maintain stability for Medicare beneficiaries in both Medicare FFS and

MA. The policies finalized for CY 2024 are projected to increase average payments to MA

organizations by 3.32% in CY 2024, which will provide continued stability to the MA market

and MA beneficiaries. The policies we are finalizing are commonsense, clinically-based

technical updates that are crucial to ensuring that payments to MA organizations are up to date

and reflect current diagnostic and expenditure trends. These updates ensure accurate payment to

MA organizations and prevent wasteful Medicare spending. These policies were proposed and

finalized using careful analyses, iterative clinical input, and with CMS’ strategic pillars,

especially our commitment to health equity, top of mind.

We respectfully disagree with some commenters’ claims that this reasonable update to payments

in MA is actually a payment cut that will result in increased costs or fewer benefits for

beneficiaries. These comments disregard the payment impact of increases in plan risk scores,

which is an essential element to understanding the full revenue picture for MA organizations.

Well-established historical trend data show that MA organizations continue to increase the

diagnostic codes they submit to CMS for payment, even when models are updated. We have

included assumptions for a reduced MA risk score trend due to the updated model. Moreover, we

note that there have been prior years when the overall expected revenue change was lower than

the 3.32% change projected in this Rate Announcement, and enrollment in MA plans and the

extra payments that MA plans receive from CMS, called rebates, to provide supplemental

benefits that are above and beyond those available in Medicare FFS, consistently increased over

the years, including the years of lower growth in payments. In fact, over the past decade MA has

become a very robust market, where plans compete for enrollment in large part by offering zero

premium plans and generous supplemental benefits. In fact, rebates have more than doubled in

the past five years and rose 20% from 2022 to 2023. The competitiveness of the MA market has

also resulted in beneficiaries having access to on average approximately 61 plans and roughly

60% of enrollees enrolled in zero premium plans.

The updates proposed by CMS in the CY 2024 Advance Notice are technical, data-driven, and

clinically-based updates that improve the accuracy of payments to MA organizations, as required

under the statute governing the MA program. We expect MA organizations that are committed to

their MA business to have strong business plans, long term financial strength, and a business

trajectory beyond a single year.

Though plans will have different impacts for the policies finalized, the robust strength and

competitiveness of the market and high level of choice ensure that the 3.32% nationally averaged

update will result in maintained stability.

Section B. Estimates of the MA and FFS Growth Percentages for 2024

Technical Update to USPCC baseline regarding MA-related Medical Education Expenses

Comment: In the CY 2024 Advance Notice, we proposed a technical change to remove medical

education costs (IME and DGME) paid to hospitals by CMS associated with services furnished

to MA enrollees from the historical and projected expenditures supporting the non-ESRD FFS

USPCCs beginning with the CY 2024 ratebook. Several commenters expressed support for the

technical update to remove medical education payments paid by CMS to hospitals associated

with services furnished to MA enrollees from the non-ESRD FFS USPCC estimates, in order to

make payments to MA organizations more appropriate and accurate.

A few commenters expressed appreciation for CMS’ ongoing efforts regarding data evaluations

and improvements regarding the development of per capita FFS costs, and for the transparency

provided by CMS concerning these efforts. One of these commenters encouraged CMS to

continue to take steps to improve the level of transparency related to the methodologies and

analysis supporting the development of the USPCCs and county benchmarks, such as providing

more details in the Advance Notice on how adjustment factors are calculated including data

source/scope and providing further explanation of changes to methodologies.

Response: We appreciate the support and feedback provided by the commenters.

Comment: A large number of commenters expressed concern regarding the full implementation

of the technical update to the USPCC baseline at one time, and recommended that the change

instead be phased in gradually over multiple years to minimize disruption to premiums and

benefits. Several of these commenters noted that CMS has previously phased in benchmark

changes, citing as examples the statutory phase-in of the Affordable Care Act benchmark

changes and the incremental IME phase-out per section 1853(k)(4)(B)(ii) of the Act.

A couple of commenters requested that the technical update be delayed and not finalized for CY

2024. An additional couple of commenters recommended that CMS not proceed with the

proposed technical update to the USPCCs for CY 2024, requesting instead that CMS provide

more information about the methodology for identifying medical education payments and the

adjustments to the county benchmarks and USPCCs. One of these commenters believed there

was insufficient time for stakeholders to analyze the potential impacts of the technical proposal.

Response: We appreciate the commenters’ suggestions to phase in or delay the technical update

to the data used for the USPCC baseline. The Secretary has directed the CMS Office of the

Actuary to phase in the technical update to the data used to develop the USPCCs over a 3-year

period beginning with 33% of this adjustment to the medical education costs applied to the

USPCCs in 2024. We expect 67% of the 2025 medical education adjustment to be applied in

2025 and 100% of the 2026 medical education adjustment to be applied in 2026.

Comment: Several commenters noted that there has already been an adjustment to remove

medical education costs from the county-level benchmarks in prior years. Many commenters

requested clarification regarding how the medical education costs being removed from the non-

ESRD FFS growth rate for the technical change is different than the medical education costs

removed from county-level benchmarks as an adjustment to the Average Geographic Adjustment

(AGA). A few of these commenters inquired whether there is any overlap between the amounts

being removed from the USPCC and those being removed at the county level.

Response: The adjustments to the USPCC and AGA pertain to two different groups of Medicare

beneficiaries: the technical update to the non-ESRD FFS USPCC pertains to excluding IME and

DGME costs associated with MA enrollees (paid directly by CMS to hospitals), whereas the

county level adjustment to the AGA pertains to IME and DGME costs associated with FFS

beneficiaries (paid directly by CMS to hospitals) to determine MA capitation rates as required by

section 1853 of the Act. Historically, IME and DGME payments included in the non-ESRD FFS

USPCCs were sourced from historical inpatient cost reports and included amounts paid on behalf

of both FFS and MA enrollees. The cost reports are used as a source for the baseline projections

of the USPCCs since the data contains more detail of the various components of hospital

payments that are projected separately, including capital, bad debt, and ancillary pass through

payments. In contrast, the IME and DGME payments used to calculate the ratebook IME and

DGME carve-out factors applied to the AGAs were sourced from the FFS claims records and, as

such, the adjustment in the county FFS rate calculation has always been limited to the payments

for FFS admissions. The claim records, and not cost reports, are used in the ratebook medical

education exclusion because the claim records include the beneficiary’s county of residence.

Therefore, no corresponding adjustment is required to the IME phase-out and DGME carve-out

adjustments to the AGAs in the county ratebook calculation to remove costs associated with MA

enrollees. As stated on page 11 of the CY 2024 Advance Notice, the technical update has no

impact on the exclusion of medical education costs from the AGAs used to develop the ratebook.

Comment: A few commenters believed that the explanation of the technical change in the

Advance Notice was limited and did not fully explain how the change will affect cost

projections, why the technical change is being implemented at this time, how the impact was

determined, and the statutory basis for the technical change.

Many commenters requested more transparency and details regarding the technical update and

the derivation of the impacts. Specific requests from commenters included the following:

• Further explanation/justification of the need for such a large adjustment, such as detailed

support of the magnitude of the impact of the technical update (including the underlying

data and calculations), information regarding how the technical update compares to the

magnitude of historical IME costs that are removed after the application of the AGA to

the rebased FFS per capita costs, and how the technical update is consistent with the

underlying county claim costs.

• Disclosure of whether CMS has performed a reconciliation of the two sources of

IME/DGME information (IME/DGME payments sourced from historical cost reports

used in the non-ESRD FFS USPCC, and IME/DGME payments sourced from the FFS

claims records used to calculate the county ratebook carve-out factors), and if so,

disclosure of the reconciliation results which could demonstrate whether the two sources

are consistent (to ensure that the FFS IME/GME payments included in the non-ESRD

FFS USPCC align with the IME/GME payments being excluded in the MA ratebook).

• Information regarding the methodology for developing the MA-related IME and DGME

adjustment to the USPCC (to better understand how CMS calculated the projected

estimates for medical education costs for 2022 and 2023), and the impact on the CY 2022

and CY 2023 non-ESRD FFS USPCCs.

• Additional information on the methodology for identifying IME and GME payments for

exclusion.

Response: CMS Office of the Actuary believes that the technical update is consistent with

projecting actuarial estimates of FFS per capita costs for the upcoming contract year, as directed

by statute. Section 1853(c)(1)(D) requires an estimate of the per capita costs for services covered

under Parts A and B for individuals who are not enrolled in an MA plan.

In the CY 2024 Advance Notice, we indicated that, for the CY 2024 ratebook development, the

baseline development and modeling supporting the USPCCs had now been updated to separately

identify the historical and projected costs of IME and DGME paid to hospitals by CMS

associated with services furnished to MA enrollees. This update in the modeling stems from

separate projections of IME and DGME by FFS versus MA coverages.

In this Rate Announcement, we are providing the updated impact of the technical update for

IME/DGME on the final estimate of the CY 2024 non-ESRD FFS USPCC (that is being released

in this Rate Announcement); the following is a demonstration of the impact.

As reflected in Table II-2 of this Rate Announcement, the CY 2024 Part B non-ESRD FFS

USPCC is $664.40. This value is unaffected by the technical update for IME/DGME.

The following table illustrates the development of the CY 2024 Part A non-ESRD FFS USPCC

both pre- and post-adjustment.

Projection for Contract Year 2024

With full (100%)

implementation of

technical update

(informational)

With 33%

implementation of

technical update

for CY 2024 rates

a. FFS Enrollment (in millions)

33.16

Reimbursements (in millions)

b. Legacy approach (i.e., including MA medical

education)

$178,800.9

c. MA medical education

($11,063.6)

($3,651.0)

d. Admin loading

1.0011

e. 2024 Part A non-ESRD FFS USPCC (legacy)

e = [(b * d) / a / 12]

$449.89

f. 2024 Part A non-ESRD FFS USPCC (with

exclusion of MA medical education)

f = [(b + c) * d) / a / 12]

$422.05

$440.70

g. 2024 Part B non-ESRD FFS USPCC

$664.40

h. 2024 non-ESRD FFS USPCC (legacy)

h = e + g

$1,114.29

i. 2024 non-ESRD FFS USPCC (with exclusion of

MA medical education)

i = f + g

$1,086.45

$1,105.10

j. 2023 non-ESRD FFS USPCC from CY 2023

Rate Announcement

$1,078.63

k. CY 2024 FFS growth rate (legacy)

k = h/j – 1 (rounded to hundredth of a percent)

3.31%

l. CY 2024 FFS growth rate (with exclusion of

MA medical education)

l = i/j – 1 (rounded to hundredth of a percent)

0.73%

2.45%

m. Impact of technical update on CY 2024 FFS

growth rate

m = l - k

-2.58%

-0.86%

The resulting impact of the technical update on the MA growth rate (based on the change in the

non-ESRD Total USPCC, which includes both FFS and Part C projections) is -1.27 percent for

full (100%) implementation of the medical education change (provided for informational

purposes) and -0.42% for 33% implementation in CY 2024.

In response to a commenter’s request for a comparison of the two sources of IME/DGME

information (IME/DGME payments sourced from historical cost reports used in the non-ESRD

FFS USPCC, and IME/DGME payments sourced from the FFS claims records used to calculate

the county ratebook carve-out factors), the following table illustrates the CY 2021 incurred IME

and DGME represented in the FFS USPCC calculations and the ratebook carveouts. Note that

the data in the illustrative table below include experience for both non-ESRD and ESRD FFS

beneficiaries.

Item

CY 2021 Incurred Amount (Billions)

IME

DGME

FFS USPCCs in 2024 Rate Announcement

$6.8

$2.4

2024 Ratebook AGA Carveouts

$6.6

$2.2

USPCCs Minus Ratebook

$0.2

Note: $0.2 billion impact represents 0.06 percent of 2021 non-ESRD FFS expenditures.

The results show that the aggregate IME and DGME ratebook carveouts are similar to the

corresponding values in the FFS USPCCs. The difference is largely attributed to the different

data sources for the IME and DGME amounts. That is, the USPCC values for IME/DGME

originate from the inpatient cost reports, whereas the ratebook IME/DGME carveouts are

tabulated from the National Claim History claim records.

Comment: A few commenters expressed concern that the technical update would inappropriately

reduce the non-ESRD FFS growth rate to below the expected growth in per enrollee Medicare

costs.

Another commenter expressed concern that the reduction in the growth rate for the technical

update will exacerbate financial challenges facing not-for-profit health plans and negatively

impact value-based care providers, many of whom are paid based on the payments Medicare

Advantage plans receive from CMS.

Response: We appreciate the concerns raised by the commenters. We note that the non-ESRD

FFS USPCCs in prior ratebook years had included both IME and DGME costs paid to hospitals

on behalf of MA enrollees. Consequently, MA benchmarks had included these admission-related

costs even though CMS, and not MA organizations, had been paying these costs associated with

MA enrollees directly to hospitals. That is, the non-ESRD FFS USPCCs in prior ratebook years

had included amounts paid for IME and GME associated with services for MA enrollees, and

those are not costs for Part A and Part B services “for individuals who are not enrolled in an MA

plan” per section 1853(c)(1)(D). Under authority in sections 1853(c)(1)(D) and 1876(a)(4), the

Secretary has directed the CMS Office of the Actuary to phase in the technical update to the

USPCCs over a 3-year period beginning with the CY 2024 ratebook, with 33% of this medical

education adjustment applied to the USPCCs in 2024. We expect 67% of the 2025 medical

education adjustment to be applied in 2025 and 100% of the 2026 value to be applied in 2026.

Comment: A commenter expressed opposition to the technical update, and seemed to suggest

that the technical update could be made to the growth rates without affecting the level of MA

rates, and called into question whether the technical update was an improvement to accuracy of

the USPCC calculation.

Another commenter seemed to suggest that CMS could adjust previous years’ USPCCs in a

manner in which plans would not be “penalized” by the technical change.

Response: The technical update results in a more accurate projection of the USPCCs because the

update properly excludes medical education costs associated with inpatient services furnished to

MA enrollees from the costs for furnishing services to FFS beneficiaries; the non-ESRD FFS

USPCC reflect FFS per capita costs and should not be based on the costs associated with services

furnished to MA enrollees. The statute prescribes the general approach per section 1853(c) to

updating the USPCCs and growth rates, and section 1853 of the Act requires that FFS per capita

costs be used in developing MA rates. As discussed in the CY 2024 Advance Notice, the growth

rates are used in the calculation of MA rates, whereby the MA growth rate is used in the

calculation of the applicable amount per section 1853(k) and the non-ESRD FFS USPCC (and

implicitly the corresponding non-ESRD FFS growth rate) is used in the calculation of the

specified amount per section 1853(n). The CY 2024 FFS growth rate is calculated as the current

projection of the 2024 non-ESRD FFS USPCC (that is being released in this Rate

Announcement) divided by the prior projection of the 2023 non-ESRD FFS USPCC (that was

released in the CY 2023 Rate Announcement). The 2024 non-ESRD FFS USPCC is used in the

calculation formula for the CY 2024 MA county rates.

Comment: A couple of commenters noted that the technical update will affect MA benchmarks

in all counties, even though specific counties may have lower levels of medical education

payments.

Another commenter believed that the technical change would have a higher impact on their

urban service market that has a high concentration of academic medical centers than the national

impact that CMS provided in the CY 2024 Advance Notice.

Response: The technical update is a revision to the national monthly per capita cost (USPCC)

that is applied to all counties as follows. The non-ESRD FFS USPCC and the corresponding

non-ESRD FFS growth percentage are used uniformly in the calculation of the specified amount

(in developing the county level FFS per capita cost estimates under Section 1853(n)(2)(A) of the

Act) for all counties. The non-ESRD Total USPCC and the corresponding MA growth

percentage are applied uniformly in the calculation of the applicable amounts (per Section

1853(k)(1)) which serve as a cap on the specified amount which affects the subset of counties

where the rates are at the benchmark cap level. We provided the preliminary impacts of the

technical update in the CY 2024 Advance Notice, and we now provide the final impacts of the

technical update in this Rate Announcement, for the FFS growth rate and the MA growth rate so

that stakeholders can understand how the technical update will impact the county rates in their

plan service area.

Consolidated Appropriations Act, 2023 (CAA, 2023)

Comment: A couple of commenters requested confirmation regarding whether the impacts of the

Consolidated Appropriations Act, 2023 (P.L. 117-328) are included or excluded from the growth

rates.

Response: As indicated on page 14 of the CY 2024 Advance Notice, the USPCCs and growth

rates provided in the CY 2024 Advance Notice did not reflect the impact of CAA, 2023, given

the timing constraints of the recently enacted CAA, 2023, and the statutory timeframe for

releasing the Advance Notice. The USPCCs and growth rates provided in the CY 2024 Rate

Announcement do reflect the provisions of CAA, 2023.

Inflation Reduction Act (IRA) Medicare Part B provisions

Comment: A commenter noted that, under the IRA, inflation rebates paid by manufacturers are

remitted to the Medicare Trust Fund for Part B drugs furnished under MA and FFS. The

commenter indicated that this may create complexities in MA reimbursement assumptions. The

commenter encouraged CMS to ensure that the assumptions related to the cost of Part B

Rebatable Drugs in the MA rate setting model accounts for the total cost incurred by MA

organizations (that is, that the MA rates should reflect the Average Sales Price of a prescription

drug prior to the application of the inflationary rebate).

Also, the commenter indicated that MA plans will be required to provide beneficiaries with

reductions in cost-sharing to ensure that their cost-sharing does not exceed 20% of the net price

of the Part B Rebatable Drug after the application of the inflation rebate. The commenter

encouraged CMS to ensure that projected MA costs for Part B Rebatable Drugs should account

for these reductions in cost-sharing for beneficiaries.

Response: For 2024 MA rate development, CMS assumed that prices for Part B drugs will not

materially exceed the inflation-adjusted payment amounts under section 1847A(i) of the Act.

Therefore, no adjustments to projected Part B FFS expenditures to account for inflation rebates

for 2024 were necessary.

Further, section 1853 of the Act sets forth how the MA capitation rates and benchmarks are set

based on FFS per capita costs.

Comment: A commenter noted that the USPCCs for 2023 and thereafter reflect cost projections

related to provisions of the IRA, including exclusion from the Part B deductible for insulin when

it is furnished through durable medical equipment and a $35 cap on beneficiary cost sharing for

insulin. The commenter noted that CMS had indicated in the CY 2024 Advance Notice that these

provisions are expected to increase Part B FFS expenditures beginning with 2023, and requested

more detail on the expected increase.

Response: We estimate that the reduced cost sharing for Part B insulin associated with DME

would reduce FFS beneficiary cost sharing by roughly $20-30 million during CY 2024, which

increases Part B FFS expenditures.

COVID-19

Comment: Many commenters urged additional transparency regarding actuarial assumptions

used to calculate the growth rates pertaining to the COVID-19 pandemic. Specific requests from

commenters included the following:

• More specificity about the extent to which coverage requirements related to the COVID-

19 public health emergency are reflected in estimates of costs, utilization, and growth

rates.

• Additional information on the actuarial assumptions that underlie the accounting of

pandemic-related costs in the calculation of the FFS growth percentage.

• Additional information about how revised USPCC projections (for ex., the 2022 USPCC

restatement) are affected by the emerging experience with COVID-19.

• Detailed information as to the complete costs related to the COVID-19 pandemic and

whether these costs are included in the growth percentages, such as: the annual utilization

trend during 2020-2022, long-term costs, costs for beneficiaries dually eligible for

Medicare and Medicaid that were disproportionately impacted by the pandemic, and the

additional hospital, testing, and vaccination costs associated with COVID-19.

• Assumptions related to future COVID-19 vaccine cost and utilization.

• Any difference in how assumptions related to the COVID-19 impact were applied to the

ESRD population relative to the broader Medicare population.

Response: Several policies and legislative provisions were enacted during the public health

emergency that increased spending; notably, the 3-day inpatient stay requirement to receive

Skilled Nursing Facility (SNF) services was waived, payments for inpatient admission related to

COVID-19 were increased by 20 percent, and the use of telehealth was greatly expanded. The

public health emergency is assumed in the actuarial modeling to end in June 2023, when these

effects are assumed to be eliminated. The results would not be materially affected if the actuarial

modeling had instead assumed the public health emergency to end on the announced date of May

11, 2023, as is currently expected.

Actual Medicare FFS per capita spending has been consistently below the pre-pandemic

projections throughout the public health emergency. The following table compares combined

non-ESRD and ESRD FFS per capita spending by major provider category with what was

assumed on a pre-pandemic basis for 2020 through the third quarter of 2022 for each category.

The pre-pandemic baseline reflects utilization trend assumptions developed prior to the onset of

the COVID-19 pandemic but has been updated to reflect certain economic factors, such as actual

price updates, and certain demographic factors, such as the share of individuals enrolling in MA

plans by ESRD status.

2021 2022 2023 2024 2025

2026

-2.50% -4.00% -4.40% -4.40% -3.90%

-2.30%

Service Type

Per capita: actual / baseline

2020 2021

2022

Inpatient -6.6% -7.1% -8.7%

Outpatient -16.3% -6.5% -8.6%

Skilled Nursing Facility 2.6% -2.3% 3.1%

Home Health -15.3% -17.9% -24.5%

Physician -12.9% -3.1% -2.7%

Note: The data for 2022 include only the first three quarters.

A number of factors have contributed to this lower spending. First, the deaths from COVID-19

have contributed to a lower average morbidity for the surviving population. Using a matched-

cohort approach we studied the impact of COVID-19 deaths on historical and projected

Medicare FFS morbidity. The resulting estimate of excess morbidity on aggregate per capita

Medicare FFS spending is in the following table:

Over the last several years, a greater proportion of those dually eligible for Medicaid and

Medicare have been enrolling in MA which has decreased the average FFS per capita cost for

inpatient hospital, SNF, and home health spending. In addition, the proportion of hip and knee

replacement surgeries performed in the inpatient setting has dropped dramatically during the

public health emergency, causing a greater shift in spending from the inpatient to outpatient

setting than implicitly assumed in prior projection assumptions. These factors account for a

significant portion of the difference between actual costs and the pre-pandemic baseline.

Assumptions for COVID-19 vaccine in CY 2024 are (i) 47 percent of beneficiaries will receive a

COVID shot with 43 percent represented in Medicare FFS claims; (ii) average doses per utilizer:

1.3; and (iii) average cost per dose: $105.

We studied the impact of the COVID-19 pandemic on historical experience separately for non-

ESRD and ESRD FFS beneficiaries. Accordingly, separate COVID-19 related projection factors

were applied to the non-ESRD and ESRD baseline projections.

ESRD growth rate

Comment: A commenter expressed appreciation for improved transparency regarding the

development of ESRD Dialysis-only FFS USPCCs, further stating that the information provided

in the CY 2024 Advance Notice Attachment I was extremely helpful.

Response: We appreciate the support.

Comment: A commenter requested that CMS provide additional detail and explanation into the

significant historical restatements of the ESRD Dialysis-only FFS USPCC, in light of significant

and directionally opposite revisions of some of the USPCCs. Several other commenters noted

that the restated CY 2022 ESRD Dialysis-only FFS USPCC had a significant decrease, which

was a larger decrease than the non-ESRD FFS USPCC, and requested additional information on

the driving factors.

Response: The CY 2022 ESRD Dialysis-only FFS USPCC is lower in the CY 2024 Advance

Notice and 2024 Rate Announcement than in the 2023 Rate Announcement due to reflection of

actual incurred experience through 4

quarter 2022 in the CY 2024 Advance Notice and CY

2024 Rate Announcement, whereas the CY 2022 ESRD Dialysis-only FFS USPCC in the CY

2023 Rate Announcement was projected based on actual incurred experience through 4

quarter

2020.

Comment: A couple of commenters expressed concerns regarding the volatility of the ESRD

growth percentage, particularly in comparison to non-ESRD growth rates, and urged CMS to

provide clarifying details on the change from last year. One of these commenters expressed

concern with the lack of transparency and requested more detailed information about the

methodologies and assumptions used to calculate the ESRD growth percentage.

A couple of commenters requested that CMS publish expenditure trends for beneficiaries with

ESRD by service category, similar to the file currently published for the non-ESRD FFS

USPCCs, for CY 2022 and onward.

A commenter requested that CMS analyze and disclose the volatility in ESRD per capita costs

that may be stemming from the COVID-19 public health emergency, recent inflationary trends,

and the impact of the 21st Century Cures Act that opened the Medicare Advantage benefit to

beneficiaries with ESRD. Another commenter requested that CMS ensure that the ESRD growth

percentage adequately accounts for the increased costs of the COVID-19 pandemic on the ESRD

population.

A couple of commenters believed that the use of CY 2021 as the base year for 2024 USPCC

estimates results in distorted estimates of per capita cost and has a negative effect on Innovation

Center models that use USPCC estimates as a benchmark for payments, and further believed that

CY 2020 data would also be problematic to use due to the pandemic. One of these commenters

urged CMS to review its USPCC estimates in this context and take steps to mitigate the effect of

health care utilization during the pandemic on the ESRD USPCC estimates, particularly any

differential impact on beneficiaries with chronic kidney disease and ESRD.

A few commenters indicated that, with ESRD enrollment increasing, and with the use of a

modified version of the USPCCs serving as the basis for benchmarks under certain Innovation

Center demonstration programs, it is important for stakeholders to receive information on factors

that contribute to the USPCC development to prevent uncertainty and volatility. A couple of

commenters suggested that CMS develop policies that would address and limit the volatility in

the ESRD USPCC and the impact of prior year adjustments, such as limiting prior year

adjustments to no more than 3 percent or spreading downward adjustments greater than 3 percent

over multiple years similar to policies adopted for the Medicare FFS program (e.g., SNF, ESRD

PPS).

Response: The ESRD dialysis USPCCs are derived from the total ESRD USPCC baseline, but

are adjusted for recent trend differences between the total ESRD and dialysis ESRD populations.

Thus, the ESRD dialysis USPCCs are projected using a base year USPCC, CY 2021 for the 2024

dialysis ESRD ratebook, trended from 2021 to 2024 using total ESRD growth with an

“adjustment factor for dialysis only.” The utilization and intensity assumptions supporting the

ESRD trends are based on multiple years of historical experience. The applicable trends are

found in the Attachment II table, “Basis for ESRD Dialysis-only FFS USPCC Trend.”

As discussed in past Rate Announcements, we believe it is important to update the FFS per

capita cost estimates using the most current FFS data available at the time those values are

announced and apply repricing adjustments to reflect changes in FFS payment rules. Similar to

prior Rate Announcements, the method for calculating the county-level non-ESRD rates and the

state-level ESRD rates includes AGAs based on a five-year rolling average of historical claims

experience, which provides some measure of stability in the rates.

The published 2022-2024 “Medicare Unit Cost Increases” by service category (available at

https://www.cms.gov/files/document/ffs-trends-2022-2024.pdf) apply to provider payments for

both ESRD and non-ESRD beneficiaries. Starting with the 2024 Rate Announcement posting, we

will add trends for the ESRD Prospective Payment System (ESRD PPS) base rate.

Comment: A commenter acknowledged that the ESRD PPS policies are outside the scope of the

Advance Notice, but a couple of commenters requested that CMS consider the interactions and

downstream effects on MA ESRD rates of payment systems when developing ESRD PPS

policies and processes, including the Transitional Drug Add-on Payment Adjustment (TDAPA)

and Transitional Add-On Payment Adjustment for New and Innovative Equipment and Supplies

(TPNIES) under the ESRD PPS.

Response: We appreciate the feedback. CMS believes the current methodology for calculating

MA ESRD rates account for products that receive the TDAPA or TPNIES under the ESRD PPS.

The CY 2024 ESRD dialysis-only FFS USPCC reflects our best estimate of the national per-

capita cost, including changes to the ESRD PPS bundled payments for variables such as payment

adjustments to the ESRD PPS base rate, including the TDAPA for certain renal dialysis drugs

and biological products and the TPNIES for certain renal dialysis equipment and supplies.

Other Comments on USPCCs and Growth Rates

Comment: A commenter noted that the non-ESRD FFS USPCC estimates for prior years in

Table I-4 of the CY 2024 Advance Notice decreased in comparison to the estimates in the CY

2023 Rate Announcement. The commenter requested additional information about the factors

contributing to the revised projections (beyond the technical update that was proposed in the

Advance Notice for medical education payments), particularly for the 2022 USPCC restatement.

Response: The CY 2022 non-ESRD FFS USPCC is lower in the CY 2024 Advance Notice and

CY 2024 Rate Announcement due to reflection of actual incurred experience through 3rd quarter

2022 in the CY 2024 Advance Notice and through 4th quarter 2022 in the CY 2024 Rate

Announcement, whereas the CY 2022 non-ESRD FFS USPCC in the CY 2023 Rate

Announcement was projected based on experience through 3rd quarter 2021. The actual

experience for 2022 is lower than projected in the CY 2023 Rate Announcement.

Comment: Several commenters expressed concern regarding the level of transparency of the

analysis and assumptions used to calculate the growth percentages, and urged additional

transparency. Specific requests from commenters included the following:

• Detailed information about the factors and assumptions used to calculate the growth

percentages, including additional details on utilization changes and unit costs by type of

service.

• More clarity about the extent to which CMS accounted for finalized payment rates for

Medicare FFS inpatient (IPPS) and outpatient (OPPS) systems, including details on

utilization changes and unit costs by type of service; and if not, reasoning as to why these

payment rates were not taken into account.

• More clarity as to the reasons why the 2024 growth rate estimates are markedly lower

than in prior years, beyond the technical change to medical education costs.

• Any analysis, explanation, and methodologies that the agency utilized and relied upon.

Response: We discussed in the CY 2024 Advance Notice the methodology, sources of data,

assumptions, and trends underlying the MA capitation rates at a level of detail consistent with

past practice. In addition to the information provided in the CY 2024 Advance Notice, CMS also

shared information about actuarial assumptions related to growth rates in its Actuarial User

Group call on February 23, 2023. Participants of the call were invited to ask questions about

assumptions supporting the CY 2024 Advance Notice growth rates. This call was widely

attended by stakeholders, and the call’s agenda and materials are available at

https://www.cms.gov/files/document/february-2023-actuarial-user-group-call-agenda.pdf.

In support of the MA ratebook growth rates, CMS has, as required under section 1853(b)(3),

included an explanation of the assumptions and changes in methodology used in the CY 2024

Rate Announcement; see the key economic assumptions underlying the USPCCs included in

Attachment II of this Rate Announcement. Consistent with prior years, with this Rate

Announcement we have published additional information regarding trends for the prior five

years and unit cost increases to the contract year at https://www.cms.gov/Medicare/Health-

Plans/MedicareAdvtgSpecRateStats/FFS-Trends.html.

Additionally, the USPCC projections reflect payment levels based on the most recent Medicare

final regulations for fiscal year 2023 or calendar year 2023.

Further, the 2024 growth rates are lower than in recent years primarily due to actual experience

for 2022 is lower than projected in the 2023 Rate Announcement. The 2024 growth rates also

reflect updated modeling to account for the effects of COVID-19 and other programmatic and

demographic changes, lower morbidity from excess COVID-related deaths, lower total spending

by explicitly modeling the shift of hip and knee replacements from inpatient to outpatient setting,

and updated modeling of the effect of a greater share of dual beneficiaries enrolling in MA.

We believe that this information in the CY 2024 Advance Notice and now this Rate

Announcement provides the necessary support for understanding USPCC levels and trends.

Comment: Several commenters expressed concern that the CY 2024 growth rate estimates were

lower than estimates in the 2022 Medicare Trustees Report, while one of these commenters

acknowledged that the calculations are not inherently performed in the same manner.

Many commenters noted that the proposed CY 2024 growth rates were lower than those finalized

in the CY 2023 Rate Announcement. Commenters believed that the proposed 2024 growth rates

did not fully account for the projected growth in costs. Several commenters encouraged CMS to

finalize a strong FFS growth percentage that fully takes into account the rising cost of care due to

inflation and provider reimbursements, and to appropriately reflect the expected increase in

Medicare costs for 2024. Commenters cited published analyses (e.g., Congressional Budget

Office, Bureau of Labor Statistics, Federal Reserve Bank of Dallas) regarding estimates of the

rate of inflation in the health care sector.

A couple of commenters expressed concern that the development of the MA rates may not fully

account for the range of costs that should be included (e.g. large operational costs, inflation,

labor attrition) that health systems are incurring, especially given that the IME/DGME technical

update reduces the non-ESRD FFS growth rate.

A commenter expressed concern that the growth rate applied to PACE rates failed to keep PACE

payment on track with the rate of inflation for medical care and other associated costs, such as

those related to workforce.

Response: The USPCC modeling approach reflects projected increases in Medicare payment rate

update factors. The projected expenditures for some of the Medicare payment systems include

the expectation of inflation including projected market basket increases for inpatient, SNF, home

health agency, and outpatient hospital projections and consumer price index (CPI) updates for

durable medical equipment projections.

The growth percentages are based on CMS’ best estimate of historical program experience and

projected trend at the time those values are announced. We continue to consider it best practice

to base the growth rates on the most recent data and assumptions available at the time those

values are announced. Therefore, for each release of the growth rates, CMS updates historical

enrollment and claims, as well as projection factors, based on the most recent data. For example,

the projections supporting the 2024 growth rates are based on actual experience through

December 31, 2022 and reflect provisions of final 2023 FFS payment regulations.

The baseline supporting the USPCCs and growth rates has been revised since the CY 2024

Advance Notice. A key change since the CY 2024 Advance Notice has been the addition of

experience for 4

quarter 2022 for the non-ESRD projections. Further, the projection factors

have been revised for the CY 2024 Rate Announcement to reflect lower morbidity stemming

from excess-COVID deaths, explicit modeling of a shift of hip and knee replacements from

inpatient to outpatient setting, and reflection of a greater share of dual eligible beneficiaries

enrolling in MA.

Section C. MA Benchmark, Quality Bonus Payments, and Rebate

Comment: A few commenters expressed concern that the cap on benchmarks imposed by the

Affordable Care Act limits health plans’ ability to improve coverage for enrollees such as adding

supplemental benefits and reducing cost sharing. A couple of these commenters encouraged

CMS to consider the impact of the benchmark cap on the Administration goal to support health

equity. A commenter stated that the benchmark cap undermines the Quality Bonus Payment

(QBP), whereby high-quality MA plans rated 4-Stars or higher will not receive the full QBP due

to the benchmark cap.

Several commenters suggested that we review our options for exercising discretionary authority

to eliminate the cap or to remove quality bonuses from the cap calculation. A couple of these

commenters referred to legal analyses provided to CMS in previous years regarding this issue

that showed that they believed such changes were legally permissible. A couple of these

commenters expressed concern that the cap is inconsistent with Congressional intent, is contrary

to ensuring that seniors receive the highest possible quality of care, and harms beneficiaries by

undermining value-based care and reducing benefits to enrollees in high quality plans.

Response: As we have stated in response to similar comments in prior Rate Announcements,

while we appreciate the commenters’ concerns, we have not identified discretion under Section

1853(n)(4) of the Act to eliminate application of the pre-Patient Protection and Affordable Care

Act (ACA) (Pub. L. 111-148) rate cap or exclude the bonus payment from the cap calculation.

Comment: A commenter expressed support for the QBP policies. Another commenter expressed

concerns regarding quality bonuses and rebates, and suggested that MA rebates be phased out.

And another commenter suggested that CMS explore changes to the rebate percentage, such as

reducing the rebate percentage in relation to increasing levels of a plan’s coding of diagnoses for

risk adjustment.

A commenter expressed concern regarding the high proportion of plans receiving quality bonus

payments based on the current star rating system. Another commenter expressed concern about

the proportion of Special Needs Plans (SNPs) that had reduced quality bonus payments resulting

from a drop in star rating under the current star rating system.

Response: We appreciate the feedback submitted by the commenters regarding quality bonus

payments and MA rebates. The statutory requirements regarding quality bonus payments and the

rebates percentages are prescribed in Sections 1853(o) and 1854(b)(1)(C) of the Act,

respectively.

Comment: A commenter supported CMS’ interpretation of Sections 1853(o)(3)(B) and

1853(c)(1)(B) of the Act with regard to Puerto Rico counties that would have had an urban floor

county rate, whereby more counties in Puerto Rico will continue to qualify for a double bonus.

Response: We appreciate the support.

Comment: A commenter suggested that, for dually eligible beneficiaries enrolled in D-SNPs in

Puerto Rico, the Part B premium buy-downs should be considered part of the A/B bid and not

considered a supplemental benefit, since dually eligible beneficiaries in the mainland would have

the Part B premium covered by Medicaid.

Response: Section 1854 of the Act specifies the costs that may be included in the bid submitted

by each MA organization. Per section 1854(a)(6)(A)(ii), the bid must separately address the costs

attributable to provision of benefits under the Medicare FFS program (as defined in section

1852(a)(1)(B), which is benefits under Parts A and B, excluding hospice and the costs of

acquisition of a kidney for transplant), including, for plan year 2020 and subsequent plan years,

the provision of additional telehealth benefits as described in section 1852(m) (that is, the “A/B

bid”) from costs attributable for supplemental benefits and Part D benefits. Payment of the Part B

premium is not a benefit under Medicare Part A or B. CMS does not have discretion under

section 1854(a)(6)(A) to treat the payment of the Part B premium as a benefit under the Medicare

FFS.

Section D. Calculation of Fee-for-Service Costs

Comment: A commenter expressed support for moving to alternative payment models that

include comprehensive prospective payment to sufficiently and sustainably support primary

care’s role in addressing patient’s health related social needs.

Response: We appreciate the feedback from the commenter about the scope of CMMI models.

Comment: A commenter expressed support for the repricing refinements applied to the

development of FFS costs, such as reflecting changes in FFS payment rules.

Another commenter expressed support for the AGA methodology, and further suggested that the

CMS-HCC risk adjustment model proposed for CY 2024 should be used for the purpose of

standardizing the AGA factors regardless of the risk model finalized for CY 2024 payment.

Another commenter suggested that CMS adjust county-level benchmarks for states and regions

that are disproportionately affected by risk model revisions.

Response: We appreciate the support and suggestions from the commenters. As in prior years,

the benchmarks will be standardized with the risk scores calculated using the CMS-HCC risk

adjustment model(s) being used for the payment year (in this case CY 2024), including any

blending/phasing-in of the risk adjustment models.

Comment: A couple of commenters urged additional transparency regarding the rebasing

methodology, given the regional variations in pandemic impacts and to ensure accuracy and

stability. A couple of other commenters suggested that CMS release a preliminary estimate of the

impact of rebasing the county rates at the time of the Advance Notice.

Response: We appreciate the request for transparency and believe that we have been responsive

to stakeholders’ interest in understanding and analyzing the rebasing methodology. As noted on

page 25 of the CY 2024 Advance Notice, CMS released the 2021 FFS cost data by county used

for rebasing county rates in the development of the 2024 ratebook. This data is available on the

CMS website at https://www.cms.gov/Medicare/Health-

Plans/MedicareAdvtgSpecRateStats/FFS-Data.html. Due to timing constraints, this data did not

reflect adjustments for Innovation Center models and demonstrations and the Medicare Shared

Savings Program and Advanced Alternative Payment Models, and do not reflect adjustments for

claim repricing for the most current available Medicare FFS payment rules and parameters.

Starting with the CY 2020 Advance Notice, CMS has published with each Advance Notice the

latest FFS cost data by county used in the development of the non-ESRD ratebooks. For the CY

2019 Advance Notice and prior, this FFS cost data was released at the same time as the Rate

Announcement on the CMS webpage at: https://www.cms.gov/Medicare/Health-

Plans/MedicareAdvtgSpecRateStats/FFS-Data. The accelerated release of the FFS experience

allows stakeholders to conduct basic analyses of the impact of recent program experience on the

geographic adjustments supporting the rates.

Comment: A couple of commenters expressed support for the use of five years of FFS

experience to mitigate any annual fluctuations and anomalies in the data.

Several commenters expressed concerns regarding rebasing the rates for CY 2024 pertaining to

the potential for instability and disparities across counties, such as in Puerto Rico and Florida.

Commenters requested that CMS consider a variety of alternatives, including: an exemption

from rebasing for Puerto Rico given the significant impact of natural disasters and the ongoing

pressures related to COVID-19, not rebasing the MA rates for CY 2024 to improve stability in

the MA program, removal of the 2020 FFS experience from the five-year period used in the

AGA calculation due to the impact of the COVID-19 pandemic, and a suggestion to make an

upward adjustment to the AGA methodology to account for the downward pressure of COVID-

19 impacts on FFS medical costs in 2020 due to suppressed utilization and the impact of natural

disasters which are occurring with greater frequency and severity in recent years.

A couple of commenters encouraged CMS to consider rebasing less frequently (e.g., AGA

update every three years).

Response: We appreciate the feedback submitted by the commenters and appreciate their

concerns about stability in MA county rates.

We note that the impact of rebasing and repricing the historical FFS data had a positive impact

on CY 2024 MA rates in Puerto Rico.

The CY 2020 Advance Notice (page 21) and Rate Announcement (pages 27 and 28) included

discussion and analysis of trends in the FFS data, and concluded that our methodology of using

five years of FFS experience mitigates annual fluctuations and anomalies in the data that may

occur for a variety of reasons. The CY 2023 Advance Notice (pages 24 and 25) also discussed

CMS’ analysis of the trends in the 2020 FFS data that were impacted by the COVID-19

pandemic and affirmed our conclusion that using five years of historical data provides for

stability in the rates despite local or regional events, such as natural or weather-related disasters,

and varying impacts from nationwide events, such as pandemics.

As discussed on page 18 of the CY 2024 Advance Notice, section 1853(c)(1)(D)(ii) of the Act

requires CMS to rebase the county FFS per capita costs periodically, which entails updating the

estimate of each county’s FFS costs using more current FFS claims information. As discussed in

past Rate Announcements, given that MA county rates are based on FFS costs, we believe it is

important to update the FFS per capita cost estimates using the most current FFS data available

and apply repricing adjustments to reflect changes in FFS payment rules. We have stated in

previous Rate Announcements that we anticipate rebasing the rates each year. We have also

previously discussed how the method for calculating the MA county rates includes a five-year

rolling average of historical FFS claims experience, which provides a measure of stability in the

rates. We are finalizing the proposal to rebase the CY 2024 rates.

Comment: A commenter indicated that they expect the impact of rebasing county FFS rates for

CY 2024 to result in a much larger, negative impact in Florida than in most other states, and

requested additional detail regarding the drivers of such a negative trend in Florida and the

magnitude of the drivers.

Response: As noted on page 25 of the CY 2024 Advance Notice, CMS released the 2021 FFS

cost data by county used for rebasing county rates in the development of the 2024 ratebook along

with the Advance Notice. This data is available on the CMS website at

https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/FFS-Data.html.

With the Rate Announcement, CMS annually provides a tool and corresponding glossary,

Medicare FFS county 20YY web.xlsm, which provides stakeholders with means to replicate the

FFS rate development. This file is available on the CMS webpage at:

https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Ratebooks-and-

Supporting-Data. Additionally, CMS has included in the CY 2021 and CY 2022 Rate

Announcements a detailed discussion of the rebasing impacts in Florida. Using these two

documents, in conjunction with the spreadsheet Medicare FFS county 20YY web.xlsm,

stakeholders are able to analyze the drivers of changes in FFS per capita costs for specific

counties from one ratebook to another.

Additionally, weighted by September 2022 MA enrollment by county, the overall AGAs

supporting the 2024 non-ESRD, non-PACE rates in Florida changed by less than -0.25 percent

relative to the same measure for the 2023 rates. Therefore, the impact of rebasing for 2024 has a

minimal impact on Florida rates.

Comment: One commenter expressed concern with our proposal to limit our adjustment of the

AGAs for Innovation Center payment and service delivery models to those listed in Table II-B3-

1 of the CY 2024 Advance Notice, and with the proposed exclusion of certain payments under

those models (e.g., care management fees) that are funded through the Innovation Center rather

than the Medicare Part A or B Trust Funds. The commenter inquired about the statutory basis for

excluding these costs from the calculation of MA benchmarks, and was concerned that as APMs

expand in scope that a growing share of FFS spending may be excluded from MA benchmarks.

The commenter expressed concern that the benchmarks may not reflect the cost of providing the

FFS benefit, and that the current approach fails to adequately determine the cost of providing a

benefit to MA enrollees that is comparable to the cost of providing the benefit under FFS.

The commenter is particularly concerned about the exclusion of advance payment of shared

savings and additional reconciliation payments paid to providers under Innovation Center

models. The commenter requested that CMS reconsider policies that exclude Innovation Center

costs and that CMS provide stakeholders with the amounts currently being excluded from the

development of FFS costs.

Response: As explained on pages 29-30 of the CY 2024 Advance Notice, we considered

adjusting the FFS claims experience for care management fees, per-beneficiary-per-month fees,

and/or advance payment of shared savings paid using the Innovation Center appropriation

instead of the Medicare Part A or B Trust Funds for other Innovation Center models conducted

in the 2017–2021 period. However, in continuing prior policy, we will not take fees of this type

into account in our adjustments to historical FFS experience when they were not funded under

Medicare Part A or B Trust Funds.

As we discussed on page 20 of the CY 2018 Advance Notice, the fees paid from administrative

accounts authorized by section 1115A of the Act are not from the Parts A and B Trust Funds,

from which Medicare claims are disbursed, so we do not consider those payments to be part of

FFS costs. Per section 1853(c)(1)(D)(i) and (n)(2)(F) of the Act, CMS uses the “adjusted average

per capita cost for the year involved, determined under section 1876(a)(4) [of the Act]” as the

base payment amount for setting MA rates. Section 1876(a)(4) indicates that FFS costs used for

MA rates are based on the estimated amount that would be payable for services covered under

Parts A and B, and types of expenses otherwise reimbursable under Parts A and B (including

administrative costs incurred by organizations described in sections 1816 and 1842). As these

costs described in section 1876(a)(4) of the Act are paid from the Trust Funds, excluding costs

paid from another appropriation is appropriate to determine FFS costs. See also sections 1817

and 1841 of the Act. In addition, section 1853(f) of the Act indicates that payments to MA

organizations shall be made from the Trust Funds “in such proportion as the Secretary

determines reflects the relative weight that benefits under Part A and under Part B represents of

the actuarial value of the total benefits under this title.” Therefore, we will not make an

adjustment to historical FFS claims to account for payments made from the funds appropriated

under section 1115A(f).

Comment: Many commenters requested that we calculate FFS spending using only claims and

utilization data for beneficiaries enrolled in both Part A and Part B (rather than based on such

data for beneficiaries in Part A and/or Part B, as is the practice today), because they believed that

would be a more accurate, reasonable, appropriate, and/or equitable methodology. Several of

these commenters cited MedPAC’s support of benchmarks calculated based on FFS data for

beneficiaries with both Part A and Part B.

Several commenters pointed out that, in order to enroll in an MA plan, beneficiaries are required

to be enrolled in both Part A and Part B, and believe that the benchmark calculations should

align with the population of beneficiaries eligible to enroll in MA plans. One of these

commenters believes the current methodology is inappropriate from an actuarial perspective, as

the current methodology includes beneficiaries who are not eligible to enroll in MA, and stated

that actuarial principles require that an estimate of the benchmark must represent what the MA

enrollee would cost in FFS. Further, the commenter believes the Social Security Act requires that

Part A-only enrollees be excluded from the calculation of county benchmarks to ensure that the

estimate best represents what that enrollee would cost in FFS.

A commenter suggested that the current approach fails to adequately determine the cost of

providing a benefit to MA enrollees that is comparable to the cost of providing the benefit under

FFS. Several commenters expressed concern that, as the number of Medicare beneficiaries with

Part A-only grows, MA benchmarks may be distorted as artificially low and fail to reflect the

FFS costs of the population eligible to enroll in Medicare Advantage, which the commenters

believe results in an actuarially inaccurate and inequivalent benchmarks. One commenter noted

that Part A-only enrollment varies by county, whereby certain counties are disproportionately

impacted.

A commenter indicated that the risk adjustment models are calibrated with FFS beneficiaries

enrolled in Part A and Part B, and recommended that risk adjustment and payment rates be based

on the same population.

A few commenters noted that, in a recently released public use file containing information on

geographic variation in Medicare spending, CMS Office of Enterprise Data and Analytics

(OEDA) excluded beneficiaries who were Part A only or Part B only and the document stated the

following regarding per-capita spending for beneficiaries enrolled in Part A only or Part B only:

“Since those beneficiaries are enrolled in only one part of Medicare, per-capita spending for

those beneficiaries cannot be compared directly to spending for beneficiaries that are enrolled in

both Part A and Part B.” A couple of these commenters interpreted this statement and the public

use file methodology to be a “tacit acknowledgement” by CMS that the MA benchmark formula

was incorrect.

A commenter noted that in 2021, CMS had indicated that the agency intended to issue a Request

For Information (RFI) on the topic of revising MA rates to be based on data from beneficiaries

with both Part A and Part B, but no such RFI has been released and the commenter encouraged

CMS to issue an RFI to gather stakeholder input.

A couple of commenters stated that, similar to their belief that the adjustment made to per capita

costs for Medicare beneficiaries who are dually eligible for benefits through the Department of

Veterans Affairs and the Department of Defense (i.e., the VA/DoD adjustment) is needed

because these beneficiaries are not enrolled in MA, a similar adjustment should be made for Part

A-only and Part B-only beneficiaries who are not enrolled in MA (because they are not eligible

to enroll).

Several commenters expressed support for continuing our policy of basing benchmarks in Puerto

Rico on Medicare costs for beneficiaries with both Part A and Part B coverage. A few

commenters requested that we apply a uniform approach in all counties to calculate benchmarks,

pointing to the methodology used by CMS for Puerto Rico rates, to improve payment accuracy

by addressing high MA penetration rates and low FFS Part B enrollment in other areas such as

Hawaii. A couple of commenters suggested that CMS could implement a phased-in approach for

counties with MA penetration over a certain percentage and gradually lower the threshold each

year.

A commenter requested that CMS revise the benchmark methodology for counties in Maryland,

to be based on the FFS experience for beneficiaries enrolled in both Part A and Part B similar to

the rate adjustment for Puerto Rico, due to the unique impact of the Total Cost of Care (TCOC)

Model in establishing benchmarks for Maryland. The commenter indicated that Medicare FFS

spending under the TCOC Model is higher than it would otherwise be under typical Medicare

FFS payment rates (e.g., IPPS/OPPS) which results in most Maryland rates being adjusted

downward by a 95% applicable percentage. The commenter indicated that revising the

benchmark methodology as they suggest would incentivize the expansion of MA offerings and

supplemental benefits in Maryland, where MA penetration is currently lower than other areas,

whereby MA plans could then be able to bid further below the benchmarks and offer more

generous supplemental benefits. The commenter noted that CMS has explicitly determined it has

authority to exercise “discretion for the data used to develop the estimate for one geographic

area, based on circumstances unique to that area” for the rates in Puerto Rico and urged CMS to

use this authority to address the unique circumstances present in Maryland to improve equity and

access to care for Maryland beneficiaries.

Response: We refer commenters to the detailed response that we provided in the CY 2020 Rate

Announcement regarding use of FFS data for costs of all Medicare beneficiaries, whereby CMS

concluded that it finds the current ratebook methodology (our longstanding policy of considering

costs of beneficiaries with Part A and/or Part B) to be consistent with the statute at Section

1853(c)(1)(D) of the Act. We continue to believe that it is not necessary to change the

methodology at this time, nor is it required as the statutory language clearly permits CMS to

include Medicare beneficiaries who have Part A only or Part B only. While we recognize that

calculating rates based on data that excludes beneficiaries entitled only to Part A would yield

different results than calculating rates based on our current methodology, that fact alone does not

determine which methodology should be employed.

With respect to Puerto Rico, we have discussed in past Advance Notices and Rate

Announcements that while most Medicare beneficiaries are automatically enrolled in Part B and

must opt out to decline it, beneficiaries in Puerto Rico must take affirmative action to opt in to

Part B coverage. As a result, we believe it is appropriate to adjust the FFS rate calculation for

Puerto Rico used to determine MA rates so that it is based only on the Medicare costs for

beneficiaries with both Part A and Part B.

Further, section 1853(c)(1)(D)(iii) of the Act explicitly requires an adjustment to the estimate of

the FFS per capita cost for individuals dually eligible for benefits through the Department of

Veterans Affairs and the Department of Defense. There is no statutory requirement for excluding

cost data for beneficiaries with coverage for Part A only or Part B only from the information

used to develop the FFS per capita cost estimate.

We appreciate the commenter’s suggestion to revise MA rates in Maryland due to the impact of

the TCOC Model on the FFS experience; however, the statute requires that MA capitation rates

be based on FFS per capita costs, not what FFS per capita costs would be under different

(hypothetical) circumstances.

The purpose of the public use file published by OEDA was to “support further analysis” of the

issue of geographic variation in the amount and quality of health care services that Medicare

beneficiaries receive. The statement cited from OEDA’s document (regarding comparing the per-

capita spending of beneficiaries who were Part A only or Part B only with per-capita spending of

beneficiaries enrolled in both Part A and Part B) was noting that beneficiaries enrolled in only

one part of Medicare have different levels of per-capita spending than beneficiaries that are

enrolled in both Part A and Part B. As noted above, while we recognize that calculating rates

based on data that excludes beneficiaries entitled only to Part A would yield different results than

calculating rates based on our current methodology, that fact alone does not determine which

methodology should be employed.

We appreciate the suggestions submitted by commenters, and we will continue to analyze this

issue and consider whether any adjustments to the methodology on this point may be warranted

in future years. For CY 2024 we will continue to calculate FFS spending for the purpose of

establishing MA benchmarks using FFS claims and utilization data for beneficiaries in Part A

and/or Part B.

Comment: The CY 2024 Advance Notice sought public comment on the possibility of adjusting

FFS experience in Puerto Rico to reflect the propensity of zero-dollar beneficiaries nationwide.

Several commenters supported the use of an adjustment to the Puerto Rico MA rates to reflect

the prevalence of zero-dollar beneficiaries nationwide. Commenters believed that such an

adjustment is appropriate because the number of zero claimants in the Puerto Rico FFS

population is a significantly greater proportion of the population relative to the rest of the United

States. One of these commenters stated that the zero-claims adjustment is needed to ensure that

plans in Puerto Rico can maintain benefits for the low-income populations they serve.

Response: The Secretary has directed OACT to adjust the FFS experience for beneficiaries in

Puerto Rico to reflect the propensity of zero-dollar beneficiaries nationwide. For purposes of

making this adjustment, consistent with the Secretary’s instructions, OACT evaluated experience

exclusively for beneficiaries that are enrolled in both Part A and Part B and are not also eligible

for VA coverage.

The updated study analyzed experience for calendar years 2017 through 2021, using the cohort

of FFS beneficiaries enrolled mid-year (i.e., enrolled in both Part A and Part B as of the mid-year

dates used for the study) to approximate the average enrollment for the year. On average, 14.5

percent of Puerto Rico FFS beneficiaries with both Part A and Part B were found to have no

Medicare claim reimbursements per year. This compares to a nationwide, non-territory

proportion of 6.1 percent of FFS beneficiaries without Medicare spending. These results were

applied to the Puerto Rico FFS experience by adjusting the weighting of the enrollment and risk

scores for the zero-claim cohort to reflect the nationwide proportion of zero-claim beneficiaries.

The resulting impact was an average increase in the standardized FFS costs in Puerto Rico of 4.4

percent for 2017 through 2021. Accordingly, a 4.4 percent adjustment was applied to the pre-

standardized Puerto Rico FFS rates supporting the CY 2024 ratebook development.

Comment: Many commenters expressed concern regarding the disparity between payment rates

in Puerto Rico and payment rates in the mainland, and urged CMS to explore all potential

options to increase MA benchmark rates in Puerto Rico. One of these commenters encouraged

CMS to work with Congressional leaders and the White House to provide a minimum level of

funding for the MA program in Puerto Rico to improve healthcare infrastructure and quality of

care, and further the goals of health equity. Another commenter urged CMS to take

administrative action to eliminate historic socioeconomic and legal disparities and long-standing

health inequities in federal programs.

A few commenters indicated that the MA rates in Puerto Rico do not adequately cover the

relatively high costs of providing care, and noted that MA plays an important role in addressing

Social Determinants of Health (SDOH) and filling in gaps in the healthcare system in Puerto

Rico. Commenters indicated that the MA program has effectively become a social safety net in

Puerto Rico, in part due to the lack of benefits that would be covered elsewhere (e.g., Part D Low

Income Subsidy).

A commenter recommended that we adjust the MA benchmarks in Puerto Rico to account for the

proportion of dually eligible beneficiaries in the Puerto Rico FFS population.

A few commenters noted that the MA program is critically important in Puerto Rico as the

foundation of Puerto Rico’s healthcare system with high MA penetration, and detailed the

socioeconomic conditions in Puerto Rico, noting the higher proportion of dually eligible

beneficiaries in Puerto Rico whereby plans devote significant resources to addressing health-

related social needs to meet non-clinical needs and to improve health outcomes. A couple of

these commenters indicated that relatively low MA rates in Puerto Rico have contributed to

provider and professional migration from the island, leading to access and quality issues.

Several commenters requested that we consider establishing a minimum AGA of 0.70 for Puerto

Rico. A few commenters noted the level of the standardized benchmarks in Puerto Rico

compared to the national level and other locales over time. A couple of commenters suggested

that a minimum AGA be implemented in a manner that is not budget neutral. One of the

commenters suggested determining a minimum AGA level similar to the geographic adjustment

under the Medicare FFS program or establishing a hold harmless minimum benchmark.

A commenter indicated their belief that a minimum AGA could be implemented by

administrative action and that CMS has the legal authority to establish a minimum AGA. The

commenter indicated that Section 1876(a)(4) does not prescribe one method to be used in

determining the adjusted average per capita cost (AAPCC) for the purposes of setting the MA

benchmark for a service area, rather the statute only directs that the Secretary estimates the

AAPCC, giving broad discretion in potential actuarial methods to be used, with a few conditions.

The commenter continued that estimating the AAPCC can be with “actual experience” of FFS

expenditures in a service area, but such expenditure data is only specified to provide “the basis”

of the Secretary’s estimate. The commenter further continued that the Secretary may forgo

“actual experience” of FFS expenditures for an alternate general method, namely an “actuarial

equivalent based on an adequate sample and other information and data.” The commenter

indicated that this broad alternate method confirms the discretion available to the Secretary and

that the AAPCC need not be merely a restatement of actual FFS expenditure experience in a

particular service area. Lastly, the commenter indicated that, whether using “actual experience”

or an “actuarial equivalent” as the basis of estimating AAPCC, the Secretary is not limited to

using expenditure data from the service area for the particular MA benchmark. The commenter

indicated that the statute expressly permits the Secretary to use data from “a geographic area

served by an eligible organization or . . . a similar area.” The commenter indicated that CMS can

select data from a similar geographic area or jurisdiction whose FFS population is expected to be

useful in projecting what the amounts payable to the plan would be for services provided to its

enrollee population in the service area. The commenter continued that it would be reasonable for

the Secretary to look to a “similar area” that is likely to predict with some accuracy what the FFS

expenditures in Puerto Rico would be if its broader population of MA enrollees received care

through FFS. The commenter indicated that a similar jurisdiction would need to be a territory

that is similarly excluded from several federal programs, such as the LIS and SSI, and depends

upon annual appropriations from Congress for Medicaid funding. The commenter continued that

a similar jurisdiction could be an island with the same relatively high costs for utilities and

imported goods and services including national prices for prescription drugs and lab services,

and that a similar jurisdiction would have a high poverty rate, high urbanization, hurricane-

damaged infrastructure. The commenter concluded that the US Virgin Islands (USVI) would

meet these criteria as a similar jurisdiction, such that the AAPCC for Puerto Rico could be

calculated by multiplying the USPCC by the USVI AGA to determine a more accurate estimated

FFS per capita spending in Puerto Rico. The commenter offered that to simplify, the Secretary

could establish a national AGA floor of 0.70 to achieve a similar result. The commenter noted

precedent under the Medicare FFS program for CMS to establish a minimum AGA (e.g., under

the Physician Fee Schedule, ESRD PPS, IPPS).

Further, the commenter noted that, given the very high MA penetration rate in Puerto Rico, the

MA rates in Puerto Rico are based on a relatively small number of FFS beneficiaries. The

commenter provided their recent analysis of FFS data in Puerto Rico which indicated that the

FFS data that is used as the basis for MA rates is not robust and is not representative of MA

enrollees in Puerto Rico, due to the proportion of beneficiaries with zero FFS claims, the

relatively small number of FFS beneficiaries, and the low proportion of dually eligible FFS

beneficiaries in Puerto Rico.

Response: We appreciate the concerns and recommendations commenters have raised regarding

Puerto Rico. We will continue to analyze these issues and consider whether any refinements to

the methodology may be warranted in future years. As discussed in the CY 2017 Advance

Notice, the law requires that MA benchmarks be based on a county’s average Medicare FFS per

capita costs, and there is no evidence that FFS costs in Puerto Rico are higher than the costs

observed in the FFS claims data and thus no basis for overhauling Puerto Rico’s MA

benchmarks. Section 1853(c)(1)(D) requires an estimate of the per capita costs for services

covered under Parts A and B for individuals who are not enrolled in an MA plan.

As we stated in the CYs 2017 and 2018 Rate Announcements, we believe that the FFS data in

Puerto Rico is sufficient for establishing accurate MA benchmarks.

In response to a commenter’s suggestion that there is precedent under the Medicare FFS program

(e.g., under the Physician Fee Schedule, ESRD PPS, IPPS) for CMS to establish a minimum

AGA, we note that these examples are based on statutory provisions that are neither applicable to

the MA program nor provide a direct analog to the provisions in section 1853 that determine the

formula for setting MA capitation rates.

Section E. Direct Graduate Medical Education

See Attachment III Section titled “Estimates of the MA and FFS Growth Percentages for 2024.”

Section F. Organ Acquisition Costs for Kidney Transplants

Comment: A commenter expressed support for the continued use of the KAC carve-out

calculation methodology.

Response: We appreciate the support.

Comment: A commenter urged CMS to closely monitor the impact of the KAC carve-out, and

suggested that CMS take steps to limit any large decreases in county benchmarks that may result.

Response: We will continue to monitor the amount of kidney acquisition costs to determine

whether refinements and improvements to the methodology for the carve-out adjustment are

warranted.

Section G. IME Phase Out

See Attachment III Section titled “Estimates of the MA and FFS Growth Percentages for 2024”.

Section H. MA ESRD Rates

Comment: The majority of commenters on this topic expressed concerns that ESRD state rates

are not sufficient to cover the cost of care for beneficiaries with ESRD. The commenters

requested that CMS re-examine the methodology to improve the adequacy, stability, and

accuracy of MA ESRD benchmarks and payment given the statutory change that allows

beneficiaries with ESRD to directly enroll in MA plans.

Commenters highlighted the potential consequences of inadequate rates, including higher out-of-

pocket costs, reduced benefits, or limited service areas as a result of the increasing enrollment of

beneficiaries with ESRD in MA plans. A commenter stated the increased costs for beneficiaries

with ESRD could increase cost sharing for all MA beneficiaries. Another commenter expressed

concerns that MA plans may limit access to care or dissuade beneficiaries with ESRD from

choosing an MA plan.

Another commenter suggested that the underlying ESRD PPS reduces reimbursements and leads

to the development of inadequate MA ESRD rates.

Response: CMS appreciates the comments regarding MA ESRD payment adequacy given the

increased enrollment into MA plans by beneficiaries with ESRD. CMS continues to analyze

these issues and consider whether, consistent with the statutory requirements for setting ESRD

rates in section 1853(a)(1)(H) of the Act, any refinements to the methodology may be warranted

in future years to ensure appropriate ESRD payment rates.

In regard to the comment that MA plans might adopt a less attractive service area or benefit

designs due to the enrollment of beneficiaries with ESRD in MA, we note that in accordance

with the beneficiary protections at section 1852(b) of the Act and § 422.110, MA organizations

are prohibited from denying, limiting, or conditioning the coverage or furnishing of benefits to

individuals eligible to enroll in an MA plan based on health status factors. Under its own

responsibilities at section 1852(b), CMS may not approve an MA plan if we determine that the

plan design and its benefits are likely to substantially discourage enrollment by certain MA

eligible individuals with the organization.

CMS will continue to monitor and investigate complaints related to plan coverage to determine if

an MA organization has designed its plan benefits in an impermissible way. If warranted, CMS

may take compliance or enforcement actions against an MA organization for failing to meet any

contract requirements, such as providing adequate access to medically necessary services.

Comment: Commenters appreciated CMS’ analysis to explore recommendations to develop MA

ESRD rates at a smaller geographic level than the state level to address geographic differences in

costs. Commenters, citing the analysis in the CY 2024 Advance Notice, questioned the potential

use of the Core-Based Statistical Areas (CBSAs) in MA ESRD rate setting given the Area

Deprivation Index (ADI) analysis. Specifically, they expressed concerns that reduced payments

for rural and medically underserved areas, relative to urban areas, may have negative impacts on

promoting health equity and supported CMS’ proposal to use statewide ESRD rates for CY

2024. The commenters also requested more information on the CYs 2023 and 2024 analyses,

what the change and impact was on ESRD payment rates in CBSAs with a low ADI compared to

a high ADI, and the number of MA beneficiaries with ESRD located in the CBSAs with a low

ADI.

However, given payment adequacy concerns, a majority of commenters maintained that CMS

should continue analyzing policy options to change how CMS calculates ESRD payment rates,

and made suggestions such as: calculating ESRD rates at a smaller geographic level, with some

commenters asking CMS to also consider adjustments to rural and medically underserved areas

to ensure beneficiary access to high-quality MA plan options. Many of these commenters stated

calculating the ESRD payment rates at a sub-state level or on a county-level basis with

adjustments may lessen financial pressures associated with delivering care to beneficiaries with

ESRD and reduce disparities among beneficiaries living in more rural and medically-

underserved areas.

A commenter recommended improving the accuracy of state dialysis rates by reflecting the

geographic distribution of MA dialysis patients. The commenter stated CMS could develop state

rates by aggregating county-level per capita costs for FFS dialysis patients weighted by county-

level MA dialysis patient enrollment, and not re-standardize the MA-weighted state rates to

constrain them to be equal to existing state rates. According to the commenter such a method

would align MA payment rates with per capita FFS costs and pay MA plans more fairly based on

the geographic distribution of MA dialysis patients.

A commenter also stated that Puerto Rico would benefit from the implementation of rate changes

based on the ADI analysis, given that ADIs in Puerto Rico are lower than the mainland. The

commenter recommended CMS implement the ADI adjustment to MA ESRD rates.

Response: We thank the commenters for their support of our analyses of sub-state ESRD rates.

Our analysis of potential changes in ESRD rates by CBSA showed that CBSAs representing the

40 percent of enrollment with the highest ADI measures (high levels of socioeconomic

deprivation), were expected to receive CY 2022 ESRD rates that were an average of 2.13 percent

lower under the CBSA-level approach. We believe our long-standing rate-setting approach is fair

and reasonable and while we will continue our use of statewide MA ESRD rates for CY 2024,

we will consider ways to conduct additional analysis into this issue.

CMS appreciates the recommendation to develop MA ESRD rates at a sub-state level. We agree

with commenters that significant changes to the current methodology should be fully examined

prior to implementation. Proposed changes to the MA ESRD rate methodology must be included

in the Advance Notice and subject to public notice and an opportunity to comment before such

changes are adopted, consistent with section 1853(b) of the Act. Consistent with our authority

under section 1853(a)(1)(H) of the Act, CMS will continue taking into consideration

commenters’ concerns and recommendations.

Comment: Commenters provided additional suggestions for revisions to the ESRD rate setting

methodology. Several commenters stated concerns that the Maximum Out-Of-Pocket (MOOP)

limit is a factor contributing to underpayment for beneficiaries with ESRD. Some commenters

believe these higher plan costs resulting from the MOOP as applied to MA beneficiaries with

ESRD will be shifted onto beneficiaries without ESRD through higher premiums for all enrollees

and reduced supplemental benefits. Commenters suggested that CMS update the MA benchmark

to incorporate the difference between FFS Medicare out-of-pocket costs and the MA MOOP to

directly increase payments for beneficiaries with ESRD.

Another commenter requested that CMS examine the possibility of applying the QBP percentage

to ESRD benchmarks. The commenter stated that currently, MA plans do not receive quality

bonus payments for coordinating care for ESRD beneficiaries. The commenter stated that the

statute provides that the quality incentive payment should be applied at the contract or plan level,

indicating that MA ESRD membership should be included when increasing benchmarks for

quality.

Response: While we appreciate the suggestions of commenters, we do not find the suggestions to

revise the ESRD rate-setting methodology to be consistent with our interpretation of section

1853 of the Act. As explained in the CY 2012 Advance Notice and CY 2012 Rate

Announcement, CMS interprets the legislative changes made by the ACA to MA payment to

indicate that all MA payment rates, including the separate rates of payment for ESRD enrollees,

should closely align with FFS Medicare costs.

As provided in section 1853(a)(1)(H) of the Act, CMS establishes separate rates of payment to

MA organizations for ESRD beneficiaries enrolled in MA plans. See also §§ 422.254 and

422.304 through 422.308. The rates used for enrollees in dialysis or transplant status are based

on statewide average FFS Medicare costs for ESRD beneficiaries in dialysis status. For enrollees

with functioning graft status, the MA county benchmark rates are the payment rates. The rates

for those in dialysis, transplant, and functioning graft status are also adjusted using a risk

adjustment methodology that is specific to the health care costs for beneficiaries with ESRD in

dialysis, transplant or functioning graft status.

We understand the concern about potential subsidization of ESRD costs by enrollees without

diagnoses of ESRD, however the data CMS uses to calculate the CY 2024 MOOP limits includes

out-of-pocket expenses from beneficiaries with and without diagnoses of ESRD because the

MOOP limits will apply to enrollees with and without diagnoses of ESRD in CY 2024. This

practice avoids discriminating against beneficiaries with diagnoses of ESRD — or any group of

beneficiaries with a particular high-cost condition or health status — that would result if there

were higher premiums, cost sharing, or MOOP amounts applicable only to those individuals with

a certain chronic condition. Additional detail on how CMS finalized MOOP limits calculations,

including the data used and the percentiles of FFS Medicare data projections that should be used

in those calculations is available in the final rule titled “Medicare Program; Maximum Out-of-

Pocket (MOOP) Limits and Service Category Cost Sharing Standards” (CMS-4190-FC4) (87 FR

22290) published April 14, 2022.

In regard to the commenter’s recommendation for CMS to add a QBP percentage to MA ESRD

benchmarks, section 1853(o) of the Act is clear that the QBP is applied to the applicable

percentage used to calculate the applicable amount under section 1853(n) of the Act, while

ESRD rates are set pursuant to section 1853(a)(1)(H) of the Act (that is, ESRD rates are not set

under subsection (n)).

Comment: Commenters expressed concerns about the market concentration of dialysis providers.

The small number of organizations in the dialysis market impacts MA organization’s ability to

negotiate reimbursement rates close to FFS Medicare rates for dialysis services. Several

commenters cited a MedPAC analysis indicating that, on average, MA contracts are paying more

than the FFS Medicare rate for dialysis treatments.

A commenter stated financial pressure from an inability to negotiate reasonable reimbursement

rates may cause some MA plans to offset higher dialysis spending by reducing costs for other

services or risk losses on beneficiaries with ESRD. The commenter stated sub-state MA ESRD

rate setting could lessen this financial pressure.

Response: CMS appreciates the feedback provided by commenters regarding this issue. Please

refer to the CY 2021 final rule titled, “Medicare Program; Contract Year 2021 Policy and

Technical Changes to the Medicare Advantage Program, Medicare Prescription Drug Benefit

Program, and Medicare Cost Plan Program” (CMS-4190-F) (85 FR 33796), which appeared in

the Federal Register on June 2, 2020, which addresses concerns regarding dialysis provider

concentration, network adequacy requirements, and the challenges MA organizations face in

negotiating reasonable reimbursement for dialysis services.

We also note that CMS is prohibited from requiring MA organizations to use particular price

structures for payment to their contracted providers. (See section 1854(a)(6)(B)(iii) of the Act

and § 422.256(a)(2).) In accordance with the beneficiary protections under section 1852(b) and

at § 422.112(a), CMS expects that MA plans will continue to ensure that their plan designs allow

for adequate access to covered services.

CMS appreciates the recommendation to develop MA ESRD rates at a sub-state level. We do not

believe that adopting sub-state MA ESRD rates will fully address the commenters concerns. We

believe our long-standing rate-setting approach is fair and reasonable and while we will continue

Refer to CMS’ Medicare Program; Maximum Out-of-Pocket (MOOP) Limits and Service Category Cost Sharing Standards

Final Rule.

Refer to CMS’ Medicare Program; Contract Year 2021 Policy and Technical Changes to the Medicare Advantage Program,

Medicare Prescription Drug Benefit Program, and Medicare Cost Plan Program Final Rule.

our use of statewide MA ESRD rates for CY 2024, we will consider ways to conduct additional

analysis into this issue.

Comment: Several commenters encouraged CMS to leverage Innovation Center authority and

value-based models to achieve greater stability in MA ESRD payments. A commenter

encouraged the Innovation Center to limit the impact of the negative trend factor on the ESRD

Seamless Care Organizations (ESCOs) in the Comprehensive ESRD Care model. Another

commenter suggested that CMS provide MA plans with the option to participate in the

Innovation Center’s kidney demonstration models.

Commenters also encouraged CMS to explore ways to modernize the Medicare conditions for

coverage and revise the definition of “dialysis facility” to provide beneficiaries with ESRD the

choice to receive care in their preferred home or in-center dialysis setting, when clinically

appropriate.

Several commenters encouraged CMS to expand the ESRD Chronic Condition-Special Needs

Plans (C-SNP) to include beneficiaries with chronic kidney disease (CKD) to slow CKD

progression.

Response: CMS notes that potential demonstrations and modernizing ESRD regulations are

outside the scope of this document.

In the CY 2024 proposed rule titled, “Medicare Program; Contract Year 2024 Policy and

Technical Changes to the Medicare Advantage Program, Medicare Prescription Drug Benefit

Program, Medicare Cost Plan Program, Medicare Parts A, B, C, and D Overpayment Provisions

of the Affordable Care Act and Programs of All-Inclusive Care for the Elderly; Health

Information Technology Standards and Implementation Specifications” (CMS-4201-P) (87 FR

79452), CMS has proposed to revise the C-SNP category named “End Stage Renal Disease

(ESRD) requiring dialysis” to be “Chronic kidney disease (CKD)” with the following conditions:

CKD requiring dialysis/ESRD, and CKD not requiring dialysis.

CMS proposed that this

particular change would not be implemented for CY 2024 in order to give CMS time to collect

data and information related to the structuring of the proposed CKD C-SNP plan bid. CMS will

provide additional bid pricing information to MA organizations if this proposal is finalized.

Comment: Commenters expressed concerns that the MA ESRD benchmarks underrepresent the

actual costs of care for ESRD beneficiaries in Puerto Rico. A commenter stated that an

insufficient number of ESRD beneficiaries remain in FFS Medicare in Puerto Rico to predict the

costs of the MA ESRD population. The commenter asked CMS to verify what percentage of the

Medicare ESRD population in Puerto Rico remain in FFS Medicare. The commenter also asked

Refer to CMS’ Medicare Program; Contract Year 2024 Policy and Technical Changes to the Medicare Advantage Program,

Medicare Prescription Drug Benefit Program, Medicare Cost Plan Program, Medicare Parts A, B, C, and D Overpayment

Provisions of the Affordable Care Act and Programs of All-Inclusive Care for the Elderly; Health Information Technology

Standards and Implementation Specifications Proposed Rule.

if CMS has established a cut-off percentage, below which FFS experience could no longer

accurately predict MA ESRD costs.

Response: CMS appreciates the concerns and suggestions that commenters have raised regarding

ESRD rates in Puerto Rico. As CMS stated in the CY 2018 Rate Announcement, CMS believes

that the FFS data in Puerto Rico is sufficient for establishing accurate MA rates and is consistent

with the statutory requirements that MA capitation rates be based on FFS costs. CMS will

continue to analyze these issues and consider whether, consistent with the statutory requirements

for setting ESRD rates in section 1853(a)(1)(H) of the Act, any refinements to the methodology

may be warranted in future years.

There is a total of 113,000 member months of Puerto Rico dialysis experience for 2017-2021

supporting the development of the 2024 dialysis ESRD rates. We believe that this level of

experience is adequate for setting ESRD dialysis payment rates in Puerto Rico.

Comment: A commenter recommended CMS make changes to the BPT so that the ESRD

subsidy falls under Medicare-covered benefits instead of under Mandatory Supplemental

benefits. The commenter encouraged CMS to make the ESRD and non-ESRD service categories

consistent, merge the ESRD and MA BPT format in the near term, and eliminate the ESRD BPT

filing in the long-term.

Response: We appreciate the suggestions submitted by the commenter related to the BPT.

Section 1853(a)(1)(H) of the Act authorizes CMS to establish “separate rates of payment” with

respect to beneficiaries with ESRD enrolled in MA plans and does not require that a competitive

bidding methodology be used for CMS capitation payments for ESRD enrollees. In setting such

separate rates, CMS has established an approach for paying MA organizations for enrollees with

ESRD that directly use the rates, rather than bids. As such, the ESRD rates are intended to be the

base rate for enrollees with ESRD, and these costs cannot be paid under the rates used in the bids

to determine payment for non-ESRD beneficiaries. Therefore, the ESRD subsidy that is

permitted in plan bids for non-ESRD beneficiaries will remain as a mandatory supplemental

benefit. MA plans (with the exception of ESRD C-SNPs) do not bid on ESRD beneficiaries. At

this time, CMS does not find it necessary to require that MA plans submit a separate A/B bid for

beneficiaries with ESRD. Regarding the commenters request that CMS eliminate the ESRD BPT

filing requirement, CMS notes that MA plans are not required to submit ESRD BPTs (with the

exception of ESRD C-SNPs).

Comment: A commenter suggested that the underlying ESRD PPS reduces reimbursements and

leads to the development of inadequate MA ESRD rates. As noted in similar comments from

prior years, the commenter reiterated the concern that MA ESRD rates are suppressed largely by

policies that inappropriately reduce reimbursement under the ESRD PPS. The commenter cited

examples such as the outlier payment pool and case-mix adjusters that they believe have resulted

in inadequate payments from ESRD PPS reimbursement that flow into the MA rate setting

process. The commenter encouraged CMS to explore policy redesigns under the ESRD PPS.

Response: CMS appreciates the feedback provided by commenters regarding the ESRD PPS and

its relationship to the development of the MA ESRD rates. Section 1853(a)(1)(H) of the Act

authorizes CMS to establish “separate rates of payment” with respect to beneficiaries with ESRD

enrolled in MA organizations. Under CMS’ authority under section 1853(a)(1)(H), and in

keeping with CMS’ interpretation of the ACA to more closely align MA payment rates with FFS

costs, the MA ESRD state rates are based on FFS costs. CMS encourages commenters to review

and respond to ESRD PPS rulemaking for the Medicare FFS program.

Comment: Some commenters urged CMS to ensure that reimbursement for new and innovative

treatments are incorporated in the MA ESRD rates in a timelier fashion. The commenters stated

that because MA ESRD rates are calculated using historical cost data, the timing may not align

with when new products will receive the Transitional Drug Add-on Payment Adjustment

(TDAPA) and the Transitional Add-on Payment Adjustment for New and Innovative Equipment

and Supplies (TPNIES) through the FFS ESRD PPS. A commenter stated MA beneficiaries have

reported difficulties obtaining access to new and innovative products. The commenters

recommended CMS develop an alternative payment mechanism for new innovative technologies

to improve access to these treatments for MA beneficiaries with ESRD.

A commenter recommended CMS pay ESRD facilities directly for the TDAPA and TPNIES

payment adjustments during the two-year payment period for MA beneficiaries, and then adjust

MA rates to account for additional FFS payment amounts once products are fully bundled.

Response: CMS appreciates the commenters’ feedback regarding adequately funding new and

innovative products for beneficiaries with ESRD. CMS believes the current methodology for

calculating MA ESRD rates accounts for payment adjustments to the ESRD PPS base rate,

including the TDAPA for certain renal dialysis drugs and biological products and the TPNIES

for certain renal dialysis equipment and supplies under the ESRD PPS. We also note that the

TPNIES and the TDAPA for certain drugs are two-year add-on payment adjustments with no

subsequent modification to the base rate. Only the TDAPA for drugs in new ESRD PPS

functional categories may result in a base rate modification. The CY 2024 dialysis-only ESRD

USPCC reflects our best estimate of the national per-capita cost, including changes to the ESRD

PPS bundled payments for variables such as payment adjustments to the ESRD PPS base rate,

including the TDAPA and TPNIES.

CMS will continue to monitor and investigate complaints related to beneficiary challenges

obtaining access to new and innovative products to determine if an MA organization has

designed its plan benefits in an impermissible way.

Section I. MA Employer Group Waiver Plans

Comment: Some commenters expressed their support for EGWPs. Commenters stated EGWPs

are an important healthcare option for many Medicare beneficiaries and the employers who have

chosen to support them in retirement. These commenters believe that EGWPs represent a

successful public-private partnership that enables businesses to give recent retirees the option to

choose a plan that provides quality care, lower premiums, and supplemental benefits. A

commenter stated EGWPs act as critical partners in providing retirees with comprehensive

Medicare benefits with enhanced value-based benefits at a lower premium than FFS Medicare.

Response: We appreciate the support.

Comment: Some commenters expressed appreciation for the inclusion of the preliminary bid-to-

benchmark ratios for EGWPs in the Advance Notice to facilitate more accurate benefit and

premium information for employers and beneficiaries.

Response: We appreciate the support.

Comment: Many commenters recommended CMS exclude negative margin plans from the

calculation of estimated bid-to-benchmark ratios for EGWPs to avoid undermining the

availability of supplemental benefits to improve health equity.

Response: As we noted in the CY 2023 Rate Announcement, we do not believe that there is a

reasonable rationale to exclude these plans because the ratios are intended to be representative of

the market. Negative margin plans are included in the non-EGWP market as well, so the bids of

such plans are included when the bid-to-benchmark ratios are developed. CMS does adjust for

factors which would otherwise result in significant differences between the EGWP and non-

EGWP market. More specifically, while the majority of plans in the EGWP market are PPO

plans, the non-EGWP market is predominantly HMO plans. EGWP individual market bid-to-

benchmark ratios are calculated separately for HMO and PPO plan types by quartile. Unlike the

HMO/PPO difference between EGWP and non-EGWP plans, there is no data to suggest that a

similar difference exists between EGWP and non-EGWP plans regarding negative margin plans

upon which CMS can judge the reasonableness of adjusting the bid-to-benchmark ratios to

account for negative margin plans.

Comment: A commenter suggested that to reduce the number of PBPs submitted, CMS should

establish a process using segment ID to facilitate additional flexibility with Part B buy-downs.

Response: As described in recent past Advance Notices and Rate Announcements, when an MA

organization submits an individual market MA plan bid to CMS, the MA organization is

permitted to use MA rebates to buy down a portion of the Part B premiums for its enrollees in

each PBP by identifying the buydown amount in the BPT as its use of the beneficiary rebate. We

then retain that rebate amount specified by the MA organization in each PBP and coordinate

directly with the Social Security Administration (SSA) to ensure that each beneficiary’s Part B

premium is appropriately calculated and takes into account the buy-down amount. Implementing

the bidding waiver as described in the Advance Notice facilitates the communication of this

information throughout CMS systems by maintaining an operational structure that is similar to

the one that exists for individual market MA plans. For this reason, we decline to make the

recommended changes, but we appreciate the commenter’s thoughts on this issue and will

continue to analyze and explore suggestions for refinements to this policy in the future.

Comment: Commenters expressed concerns that the published preliminary bid-to-benchmark

ratios are lagging inflation and lower than expected and expressed concerns that the preliminary

bid-to-benchmark ratios and risk model changes may increase costs for employers who utilize

EGWPs and result in increased premiums or reduced benefits for beneficiaries. Many requested

additional details about how the ratios were calculated.

Response: Please see CYs 2020, 2021, 2022, and 2023 Advance Notices for additional details on

the calculation of bid-to-benchmark ratios; responses to comments on those detailed explanations

are provided in the applicable Rate Announcement. CMS is publishing updated bid-to-

benchmark ratios in this CY 2024 Rate Announcement, using the methodology detailed in the

CY 2024 Advance Notice. CMS published preliminary bid-to-benchmark ratios for the first time

in the CY 2023 Advance Notice as a result of feedback from the industry on the CY 2022 bid

cycle. MA organizations indicated that having this information early provides valuable

information in their negotiations with employer/union groups to create more accurate benefit and

premium quotes for their employer/union groups. Commenters that requested additional details

on how the bid-to-benchmark ratios were calculated did not provide further information on

which details CMS could expand upon.

Comment: A commenter requested that CMS provide updated EGWP ratios based on February

enrollment data before the release of the Rate Announcement.

Response: We appreciate this recommendation. For this CY 2024 Rate Announcement, MA

EGWP ratios are based on February enrollment.

Comment: A commenter suggested adjusting current rate setting to capture differences in the use

of HMO and PPO plans between the EGWP and non-EGWP markets. The commenter believes it

would be more accurate for CMS to segment the benchmark calculation by HMO and PPO

products and adjust the bid-to-benchmark ratio for the differing products accordingly.

Response: We appreciate this suggestion; however, we are continuing to apply our current

methodology for paying EGWPs in CY 2024. Consistent with how we have developed EGWP

payments since 2019, the CY 2024 EGWP payment methodology takes into account the

prevalence of HMO and PPO enrollment in the EGWP market by calculating CY 2024

individual market bid-to-benchmark ratios separately for HMO and PPO plan types by quartile.

CMS then takes into account the prevalence of HMO and PPO enrollment in the EGWP market

to combine the ratios by quartile. This methodology is more consistent with the county rates for

individual market plans, which are also not calculated separately for HMO and PPO plan types.

Section J. CMS-HCC Risk Adjustment Model for CY 2024

Comment: Commenters stated their support for the proposed CY 2024 CMS-HCC model, stating

it improves payment accuracy, program integrity and helps address excess payments to MA

organizations that have negatively affected taxpayers and beneficiaries. Specific examples of

support included the following comments:

• Updating the model with more recent underlying data will improve the accuracy of risk

adjustment and payments to MA organizations;

• Updating the model to account for the transition from ICD-9 to ICD-10 provides more

quality diagnostic information;

• Making changes to address variation in coding will decrease the impact of variability in MA

coding compared to FFS.

Response: We thank the commenters for their support. We agree that moving to the 2024 CMS-

HCC model will improve payment accuracy to MA plans and that updating the risk adjustment

model is an essential part of CMS’ duty to effectively run the MA program and be a steward of

the Medicare program.

Comment: Many commenters requested CMS phase in the updated 2024 model over multiple

years.

Response: CMS is finalizing the updated risk adjustment model proposed in the CY 2024

Advance Notice and will phase it in over 3 years. The 3-year phase in is consistent with how

CMS has approached other years in which model updates have been phased in over time (e.g.,

the 2014 model was phased in over 3 years and the CMS-HCC model adopted in the CY 2019

Rate Announcement to address the 21

Century Cures requirements was phased in over 4 years,

with additional revisions adopted in the CY 2020 Rate Announcement). For CY 2024, risk scores

will be calculated as the sum of 33% of the risk score calculated using the updated 2024 CMS-

HCC risk adjustment model with 67% of the risk score calculated using the current 2020 CMS-

HCC risk adjustment model. For 2025, we expect risk scores will be calculated as the sum of

67% of the risk score calculated using updated 2024 CMS-HCC risk adjustment model with 33%

of the risk score calculated using the 2020 CMS-HCC risk adjustment model. For 2026, we

expect that 100% of the risk scores will be calculated using the updated model.

Comment: Commenters recommended CMS delay implementation of the CY 2024 CMS-HCC

risk adjustment model, or withdraw it, so there could be more time to assess and understand it,

and make modifications. Some commenters requested that CMS seek additional industry input

before implementing any substantive changes to the risk adjustment model. A few commenters

cited precedent such as an October 2015 memo related to the CY 2017 Advance Notice as an

example when CMS solicited early feedback from the public on substantial changes to the risk

adjustment model. Commenters requested that CMS provide at least 60-day notice for changes to

the risk adjustment model. Some commenters cited risk adjustment model changes related to the

21st Century Cures Act that required a 60-day comment period in statute. A few commenters

recommended CMS better align the timing of the Advance Notice with the bid cycle to give MA

organizations more time to analyze impacts. One of these commenters recommended CMS

provide a 60- to 90-day comment period as well as release the Advance Notice between October

and December to give industry more time to analyze changes. Another commenter recommended

that any major risk adjustment model changes should be finalized at least one full year before

plans submit bids for the relevant benefit year.

Response: CMS appreciates the concerns raised by the commenters on the timing of the

implementation of the updated risk adjustment model and the process for comments.

As described in more detail later in this section and illustrated in Table III-1, it is important to

maintain or improve the accuracy of the risk adjustment model by updating it to reflect more

recent relative costs, treatment and utilization patterns, and coding practices. As the current

model ages and is used to predict expenditures for more recent enrollees in MA plans, we

identify in our analysis that predictive accuracy begins to decline. Delaying implementation of a

risk adjustment model that is based on more recent underlying data will prolong the use of a risk

adjustment model that, though still accurate according to CMS’ measures (i.e., having a

predictive ratio between 0.90 and 1.10), is waning in its ability to predict current costs. In

addition, the longer CMS waits to update the risk adjustment model, the more impactful it will be

on the year-over-year bottom line impacts for MA organizations.

It is vital that the CMS-HCC risk adjustment model is as up-to-date as possible and that CMS is

able to implement routine updates that improve accuracy and address variation in coding that

could lead to excess payments to plans. When a risk adjustment model is updated with more

recent underlying data, the relative factors associated with each demographic factor and HCC

reflect more recent costs, coding, and diagnostic patterns. All the updates in the 2024 CMS-HCC

model are steps that CMS has taken before to update the risk adjustment model and are not novel

changes to MA payment. In fact, CMS alerted plans about its commitment to move to ICD-10 in

2018 and to update the underlying data years. CMS is committed to implementing the new 2024

risk adjustment model, given that it reflects more recent data, is built using ICD-10 codes and a

clinically-based review of the best diagnoses to accurately predict Medicare costs.

We acknowledge the commenters’ request for an extended comment period. Per section

1853(b)(2) of the Act, the Advance Notice of proposed changes to the methodology and

assumptions used to determine annual MA capitation rates and the risk and other factors used in

adjusting MA capitation rates under section 1853(a)(1)(C) is required to have a minimum 30-day

comment period. The Advance Notice was released on February 1, 2023 and comments were

accepted through 6 PM Eastern Time on Monday March 6, 2023 (33 days). The statutory

requirement for a 60-day comment period applied only to proposals to implement certain

changes to the CMS-HCC model stipulated in the 21st Century Cures Act. As added by the 21st

Century Cures Act, section 1853(a)(1)(I)(iii) required that CMS provide at least 60 days for

public review and comment of proposed changes to the Part C CMS-HCC risk adjustment model

that were specifically based on section 1853(a)(1)(I).

Like in previous years when similar changes were proposed, CMS believes that the period

provided for comments on the CY 2024 Advance Notice is sufficient. In setting these timelines,

we seek to achieve multiple goals, including providing the statutory-required amount of time for

public comment while also releasing the Advance Notice using more current data to calibrate the

model and ensuring that the Rate Announcement is published by the statutory deadline. We

provided the public with sufficient information to review the proposals since we informed the

industry that the evaluation to reclassify the model was underway as far back as 2018 and we

provided a number of resources to evaluate the updated model. Further, the updates proposed are

in line with typical model updates for which CMS has provided a similar or shorter comment

period per the existing statutory requirement at the time. Specifically–

• In the 2018 Report to Congress,

CMS stated that it was conducting analyses for the

reclassification of the CMS-HCC risk adjustment model on ICD-10. In addition, CMS

outlined key specific areas in which the reclassification was being examined (e.g., episode of

care codes). The report also provided information for how CMS solicits clinical input, and

how CMS would evaluate the conditions and underlying diagnoses for inclusion in the

payment model.

• In the 2024 Advance Notice and associated releases, CMS provided a variety of information

to support the review and evaluation of the updated risk adjustment model as well as

explained the process used to develop, and the scope of changes made as part of the updated

model. The Advance Notice included or provided a link to sufficient material to evaluate the

updated risk adjustment model: the relative factors for each variable in the updated model, a

table that detailed the differences for every HCC in the current model relative to the updated

model, and the mapping of every ICD-10 diagnosis code that maps to the current 2020 model

and the updated model. In addition, CMS publicly released risk adjustment model software

which can be used to simulate risk scores under the updated model. We also released to each

MA organization and certain demonstrations plan-segment level risk scores for their

enrollees, by every model segment, for the current model, the proposed model, and for an

alternative model that had a subset of the updates. The information provided enabled

evaluation of the impact of various model updates, including the underlying HCC

restructuring (diagnosis to HCC mappings & HCC comparison table), the relative impact of

the updates (relative factor tables), the impact of the updates (plan-segment risk scores &

Refer to Section 4. Ongoing Research in the 2018 Report to Congress.

model software), and the impact of the changes focused on Principle 10 (plan-segment risk

scores under the proposed and alternative model).

• CMS has updated the underlying data in the model many times since the initial

implementation of the CMS-HCC risk adjustment model (see the Advance Notices and Rate

Announcements for 2007, 2009, 2013, 2014, 2017, and 2019). In addition, in the 2014

Advance Notice, CMS proposed a model (the “2014 model”) with similar updates to those in

the 2024 (e.g., data year update, a clinical revision that resulted in newly built HCCs, and

updates based on a review a diagnoses codes that showed indication of variable coding and

removal of certain codes and associated HCCs).

For the recent years in which the comment

timeline was extended compared to normal practice CMS made atypical structural updates to

the CMS-HCC model. For example, the 2017 model revised the underlying segmentation

structure of the model moving from a single to six community segments to improve the

sensitivity of the model and add protective features to make unique adjustments for every

health condition based on dual-eligibility status. This change was not phased in and the new

model was implemented 100% in year 1. For the 2019 and 2020 Advance Notices, we

proposed and finalized the changes to risk adjustment that were required in the 21st Century

Cures Act, such as the requirement to add an additional increase to payment when an

individual as multiple conditions – called count variables. The 21st Century Cures Act also

mandated that we provide 60 days comment for those specific changes to MA risk

adjustment, so we released a ‘Part I’ with the risk adjustment changes to comply but released

the remainder of the Advance Notice in ‘Part II’ to ensure we had the most up to date data

and analyses possible. The 2024 CMS-HCC model is not analogous to these situations

because it maintains the same structure adopted in 2017 and 2020.

CMS engages with stakeholders on a regular basis through various lines of communication. We

have regular meetings with various stakeholders – MA organizations, provider groups,

beneficiary advocates, trade groups, expert clinicians, disease specific advocacy groups,

academic researchers, and others – conduct user group calls with the industry, and release

Requests for Information (RFIs). In July of 2022, we released a comprehensive RFI that solicited

feedback for ways to improve the MA program. In response to this RFI, we received roughly

4,000 comments. We will continue to consider additional ways in which we can engage with

stakeholders as we consider changes to the CMS-HCC risk adjustment model for future years,

and appreciate commenter input.

Regarding the comment to align the Rate Announcement with the bid cycle, section

1853(b)(1)(B) of the Act requires the Rate Announcement to be published by the first Monday in

Principle 10 -Discretionary diagnostic categories should be excluded from payment models. Diagnoses that are particularly

subject to intentional or unintentional discretionary coding variation or inappropriate coding by health plans/providers, or that are

not clinically or empirically credible as cost predictors, should not increase cost predictions. Excluding these diagnoses reduces

the sensitivity of the model to coding variation and coding proliferation.

Refer to Section G. of the 2014 Advance Notice

April before the contract year; this is roughly 2 months before the statutorily-set bid submission

date of the first Monday in June.

Comment: Some commenters asked for more information from CMS regarding the updated

model. The information requested included: underlying data used to calibrate the model, analysis

supporting the predictive accuracy of the model, assessment of the model’s impact on certain

subgroups and plan types (e.g., individuals dually eligible for Medicare and Medicaid, all types

of Special Needs Plans (SNPs), and racial and ethnic minorities), and details on the technical and

clinical expert panel review. Examples of the suggestions for analysis and/or additional

information CMS should release to the public included:

• The reclassification methodology including the underlying data used to calibrate the model

so that stakeholders may simulate the proposed model revision and provide meaningful

feedback;

• An analysis on the effect of the case-mix between MA and FFS because of concerns that

sicker beneficiaries are leaving MA for FFS and healthier beneficiaries are moving from

FFS to MA resulting in HCC coefficients that are too high;

• An analysis supporting the predictive accuracy of the model including providing predictive

ratios and r-squared (R

) values, similar to what has been provided in previous iterations;

• An analysis to support departing from a recent statement in the Report to Congress that the

2020 model was performing well;

• An analysis to determine if plan-level variations in coding of certain conditions is associated

more with differences in beneficiary composition than upcoding;

• An analysis using the most recent PY 2022 data to reevaluate model estimates;

• Software, coefficients, and diagnosis mappings for an updated model with the updated data

years and ICD-10 changes, but without the Principle 10-focused clinical updates related to

variable coding (the V27 model); and

• Supporting data for removing diagnoses and HCCs, an analysis on diabetes, and other data

justifying Principle 10 (i.e., addressing variable coding).

Response: We appreciate commenters requests. As noted in the Reports to Congress, the CMS-

HCC risk adjustment models have consistently performed well, and the latest Report to Congress

included analyses of the performance of the 2020 risk adjustment model.

CMS has updated the

underlying data in the model many times since the initial implementation of the CMS-HCC risk

adjustment model (see the Advance Notices and Rate Announcements for 2007, 2009, 2013,

2014, 2017, and 2019). Further, CMS has always made updates to model calibrations to account

for updates and changes in International Classification of Diseases, Clinical Modification

diagnosis codes. Periodically, CMS conducts a clinical revision of the CMS-HCC model and

rebuilds all the HCCs from the ground up and, in the process, takes into account changes in

2018 and 2021 Reports to Congress: https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Risk-

Adjustors-Items/ReportToCongress

disease patterns, treatment methods and their costs, and coding practices, as well as

compositional changes within the Medicare population. For example, in 2014 we implemented

an updated CMS-HCC model that included a clinical revision using newly built HCCs using

ICD-9 codes, for which coding patterns had changed in the 10 years since the initial CMS-HCC

model. For CY 2023, CMS proposed and implemented an RxHCC model with changes similar to

those made to the 2024 CMS-HCC model. The CY 2023 RxHCC model was updated based on

more recent underlying data years and the model RxHCCs were reclassified based on the ICD-10

classification system.

As noted in the CY 2024 Advance Notice, the 2024 CMS-HCC model results in more

appropriate relative weights for the HCCs included in the model. CMS weighs a number of

considerations when updating and revising the CMS-HCC model. Since MA plans bear full

financial risk for the population and services they cover, a main objective of risk adjustment is to

minimize incentives for MA plans to compete for the healthiest beneficiaries. The risk

adjustment model accomplishes this by accurately predicting relative risk across subgroups of

beneficiaries and reimbursing plans more for populations that are expected to be sicker and have

more complex health needs. This is achieved through the segmentation of the model that assigns

unique risk scores for each condition based on demographic factors. A key focus of the MA risk

adjustment model is to accurately predict costs that are attributable to characteristics that are

present over time (e.g., chronic conditions that persist or affect longer term costs, demographics,

etc.). In this sense, the model is an insurance-like model that seeks to balance the over and under

prediction errors so that the average actual expenditures for a sufficiently large group of

beneficiaries equals the average predicted expenditures.

Further, because we are using the CMS-HCC model for payment (as compared to an analytic

exercise to predict costs), we follow our longstanding principles when updating the model in

order to ensure both statistical and clinical meaningfulness.

Because of the risk adjustment model’s focus on predicting foreseeable costs, a key factor in

measuring the predictive accuracy of the model is measuring how well it predicts costs over

subgroups of beneficiaries. Predictive accuracy in the CMS-HCC models is measured by the

predictive ratio – the ratio of predicted cost to actual cost – for a group of beneficiaries. While

perfect prediction is a predictive ratio of 1.0, every subgroup does not need to have a predictive

ratio of 1.0 to indicate a high performing model. Rather, when comparing models for groups of

beneficiaries, a predictive ratio closer to 1.0 indicates a better prediction, but a predictive ratio

between 0.90 and 1.10 is generally considered accurate.

The predictive ratios for each group of

FFS beneficiaries, when divided into 10 subgroups based on level of risk (i.e., costs predicted by

the model), show that the model predicts well across the spectrum of risk, meaning that, on

https://www.cms.gov/files/document/report-congress-risk-adjustment-medicare-advantage-december-2021.pdf, p. 42

average, MA plans with varying risk profiles are compensated commensurate with their expected

experience.

By this measure, as explained in more detail below, the 2024 CMS-HCC model that we are

finalizing improves on the predictive accuracy of the 2020 CMS-HCC model.

In addition, in prior years, we either have elected or have been directed by Congress to improve

prediction for selected subgroups of beneficiaries. For example, the 2017 CMS-HCC model

established separate model segments for populations dually enrolled in Medicare and Medicaid.

The 2014 CMS-HCC model, which had a single-segment for all enrollees residing in the

community, under predicted spending for full benefit dual eligible groups in the model sample

by close to 9 percent. Based on these findings, CMS elected to develop a model that improved

prediction for these beneficiaries. Predicting cost for dually eligible beneficiaries in separate

segments improved their predictive ratio to 1.0 and subsequently improved the accuracy of

payments to MA plans, which were increasingly specializing in dual eligible beneficiaries.

Congress, through the 21st Century Cures Act, also directed CMS to make changes to the model

to address concerns about the predictive accuracy of the model for beneficiaries with multiple

chronic conditions and for specific behavioral health conditions. The addition of count variables

and clinical revisions to HCCs improved the predictive ratios. The previous changes we made to

improve the predictive accuracy of the model are maintained in the 2024 CMS-HCC model.

There are a number of reasons to update and revise the CMS-HCC model, all of which improve

or maintain predictive accuracy. We improve accuracy of the model by updating the CMS-HCC

model with more recent underlying data years to better reflect the more recent relative costs and

more recent treatment and utilization patterns for conditions in the model, thereby avoiding any

reductions in predictive accuracy that can result from using relative factors based on older data.

For example, the actual expenditure per person for beneficiaries diagnosed with Multiple

Sclerosis (HCC 77 in the 2020 CMS-HCC model) increased 24.2 percent from $17,966 in the

2014/2015 sample to $22,308 in the 2018/2019 sample. The prevalence of Multiple Sclerosis

also increased by 11.6 percent between the 2014 and 2018. Updating the model with more recent

data results in higher a relative factor for Multiple Sclerosis to better predict cost for MA

beneficiaries. While maintaining or improving the model’s predictive accuracy by decile of

predicted risk is necessary, keeping the model up-to-date with more recent experience is also

necessary. For example, as Table III-1 shows, the current model 2020 CMS-HCC model that is

calibrated using 2014 diagnoses and 2015 costs for 2019 beneficiaries, to their actual cost in

2019, by decile of predicted risk. While the current model maintains predictive accuracy, it is

less accurate than models with more recent data because of inflation, changing beneficiary

characteristics, and differences in diagnostic coding patterns in FFS. Models based on more

recent data and coding patterns are the most accurate possible representation of what experience

is expected to be in future payment years. The 2024 CMS-HCC model is more accurate than the

2020 CMS-HCC model because it reflects more recent relative cost (2019 compared to 2015)

and includes clinically meaningful conditions that predict cost developed from experience with

ICD-10.

Measures of model performance that are based on individual observations such as the R

, which

measures how well the model explains variation in spending across individuals (rather than

groups), are produced with the risk adjustment model but are not a primary factor when

determining the model’s specification. Given the model’s goal of predicting costs over

subgroups, it is not intended to predict accurately for individuals or small subgroups of

beneficiaries. As such, measures of model performance for individuals are less relevant.

Nevertheless, the 2024 model has an improved R

compared to the current 2020 model.

Table III-1 provides predictive ratios by decile of risk, meaning the higher the risk the higher the

risk score, for the current 2020 model using both a 2014 – 2015 sample population (which is

what it was calibrated on) as well as the current 2020 model using the updated 2018 – 2019

sample population (used for the updated 2024 model), and the updated 2024 model using the

2018 – 2019 sample population. The predictive ratios provided in Table III-1 are the ratio of

2019 expenditures predicted by the model for 2019 to the actual 2019 expenditures using 2018

diagnoses. We have also provided predictive ratios by demographic segments in Attachment VIII

Tables VIII-5 – 11 for multiple subpopulations. As indicated in the below table, the current

model for both 2014 – 2015 and 2018 – 2019 sample population continue to predict expenditures

accurately, having predictive ratios between 0.90 and 1.10, but we observed that over time the

accuracy begins to decline. When comparing the same 2018 – 2019 sample population, we see

that for all but one decile the updated model predicts expenditures more accurately than the

current 2020 model.

Table III-1 Predictive Ratios by Decile of Predicted Risk

2014/2015

Sample

2018/2019 Sample

Decile 2020 Model 2020 Model 2024 Model

Improvement in

Predictive Risk

Entire sample

1.000

0.974

1.000

First (lowest) decile

0.971

0.910

0.970

Second decile

0.985

0.946

0.986

Third decile

0.987

0.943

1.013

Fourth decile

0.986

0.964

1.000

Fifth decile

1.001

0.981

0.994

Sixth decile

1.003

0.977

0.996

Seventh decile

1.005

0.987

0.999

Eighth decile

1.006

0.988

0.998

Ninth decile

1.006

0.993

1.008

Tenth (highest)

1.002

0.966

1.003

Top 5%

0.998

0.948

0.999

Top 1%

0.985

0.922

0.984

Top 0.1%

0.952

0.885

0.966

NOTES:

1. “Improvement in Predictive Risk” compares the distance the predictive ratios are from 1.0 for the 2024 model and 2020

model with a 2018 – 2019 sample.

2. A green arrow indicates that the predictive ratio for any specific decile for the 2024 model is closer to 1.0 than the

predictive ratio for the 2020 model with a 2018 – 2019 sample, and vice-versa.

Table III-2 provides the R

for the CMS-HCC model segments for the current 2020 CMS-HCC

model and the 2024 CMS-HCC model. As previously stated, CMS uses the predictive ratio as

the primary measure of accuracy for the model and R

is less useful in assessing the predictive

accuracy of the MA CMS-HCC risk model. However, we provide the R

for the model segments

because it does provide information about the extent to which the model can explain variation in

expenditures between individuals. Table III-2 shows there is slight improvement in all

demographic segments, including Full-Benefit Duals. (Increases in R

values indicate

improvement.)

Table III-2 R-Squared by CMS-HCC Model and Demographic Segments

Demographic Segment

2020 Model

(14/15 Sample)

2024 Model

(18/19 Sample)

Non-Dual, Age≥65

0.1257

0.1355

Non-Dual, Age<65

0.1148

0.1472

Full-Benefit Dual, Age≥65

0.1214

0.1246

Full-Benefit Dual, Age<65

0.1317

0.1889

Demographic Segment

2020 Model

(14/15 Sample)

2024 Model

(18/19 Sample)

Partial-Benefit Dual, Age≥65

0.1122

0.1159

Partial-Benefit Dual, Age<65

0.0987

0.1589

Institutional

0.1087

0.1200

CMS released a variety of information to support the review and evaluation of the updated risk

adjustment model as well as explaining the process used to develop, and the scope of changes, in

the updated risk model. As previously noted, the Advance Notice included or provided a link to

sufficient material to evaluate the new risk model: the relative factors for each variable in the

updated model, a table that detailed the differences for every HCC in the current model relative

to the new model, and the mapping of every payment ICD-10 diagnosis code under the current

model and the updated model. Though the information released in the Advance Notice was

sufficient to analyze the model, CMS also publicly released risk adjustment model software

which could be used to more quickly simulate risk scores. We also released to each MA

organization and certain demonstrations plan-segment level risk scores for their enrollees, by

every model segment, for both the proposed model and for an alternative model that had a subset

of the updates. The information provided enabled evaluation of the impact of various model

updates, including the underlying HCC restructuring (diagnosis to HCC mappings & HCC

comparison table), the relative impact of the updates (relative factor tables), the impact of the

updates (plan-segment risk scores & model software), and the impact of the changes focused on

Principle 10 (plan-segment risk scores under the proposed and alternative model that had a

subset of the updates in the new risk model we are adopting).

As with every model update, CMS noted in the CY 2024 Advance Notice that there will be

variation in the impact on risk scores depending on the clinical profile of each plan’s enrollees.

All of the model updates (i.e., underlying data updates, denominator update, and ICD-10

reclassification) contribute to changes in the relative costs of conditions, and therefore changes to

the resulting risk scores. Beneficiary risk scores or plan average risk scores may change

depending on an individual beneficiary’s combination of diagnoses or the clinical profile of a

plan’s enrollee population. Further, the risk scores of different subgroups can change differently,

to reflect changes that may be particular to that subgroup or clinical profiles that are more

prominent in a subgroup. These subgroups include dual status, Long Term Institutional (LTI)

status, geographic location, plan type, to the extent that plans focus on enrolling beneficiaries

with different characteristics.

Risk score differences between the current model and the updated model will occur for several

reasons. Specifically, revisions to the models result in changes in the marginal cost attributable to

each HCC, relative to the change in the average cost (i.e., denominator used to set the relative

factors), and can alter the relative factor associated with each HCC, and with the relative values

among HCCs. In addition, changes in the relative factors will result from changing from HCCs

that were created using the ICD-9 classification system to HCCs that were created using the

ICD-10 classification system, as well as from the addition or deletion of HCCs to or from the

model.

Comment: Many commenters stated that the updated model will have a significant negative

effect on the availability of supplemental benefits, and specifically identified a variety of benefits

that they thought would be affected. Commenters mentioned increased beneficiary cost sharing,

and reduced supplemental benefits such as transportation, nutritious meals, dental coverage,

hearing coverage, vision coverage, telehealth coverage, fitness and wellness benefits, decreased

cost-sharing, annual vision exams and coverage for glasses or contact lenses, as the

consequences of adopting the new model.

Response: We estimate that the overall impact of the CY 2024 Rate Announcement will be a net

increase of payment of 3.32%. The updated risk adjustment model is one of many factors that

contribute to the net payment increase, and when MA organizations prepare their bids and

benefit packages, they take into account all these factors when they determine the amount of

supplemental benefits they will offer. In addition, MA organizations have a variety of business

decisions that they make when determining how much profit margin they include in their bids,

how to allocate revenue to supplemental benefits, and the amounts that they allocate to each

category of supplemental benefits. Given the competitive nature of the MA program, we

anticipate that MA organizations will strive to minimize the impact of any payment-related

changes on beneficiaries and the scope of benefits available to them. While impacts may vary by

MA plan for a variety of reasons, given the increased revenue projected for MA organizations,

we expect that the availability of supplemental benefits will be stable or grow in CY 2024, as

seen previously in years with comparable updates. Following those payment updates, the MA

market remained stable and enrollment in MA continued to grow.

Comment: Many commenters stated that the updated model will impact care delivery and the

quality of care. Many of these commenters expressed concerns regarding the number of ICD-10

codes that no longer map to HCCs in the model, as well as some of the specific clinical areas that

are not included in the new CMS-HCC risk adjustment model, and expressed concern that the

changes made in the updated model will directly affect patient care. Specifically, commenters

cited that there are over 2,000 net fewer ICD-10 codes mapped to the updated model compared

with the current model. The commenters focused on the changes to the volume of diagnoses

mappings instead of the specific clinical updates, and stated that removing these diagnosis codes

may have significant health implications, and may prevent slowing disease progression from

early stages. In addition, some commenters stated that the conditions removed from the proposed

model are precursors to adverse medical events, and that their removal would disincentivize

catching these conditions early. Some commenters stated that the recalibration of the updated

model lowers HCC coefficients which will result in negative clinical outcomes particularly for

vulnerable populations. A few commenters recommended CMS consult with all medical

specialties prior to making any clinical changes to the HCC reclassification. Commenters

expressed concern of possible unintended consequences to patient care that may occur since they

envisioned that clinician coding and documentation practices would become less precise for

early or mild stage disease in response to the model update. Commenters highlighted specific

conditions as areas of concern (which we address in specific responses below).

Response: Risk adjustment is intended to reduce or eliminate “the incentives to enroll only the

healthiest, and thus least expensive, beneficiaries while steering clear of the sickest and

costliest—thereby rewarding Medicare Advantage insurers to the extent that they achieve

genuine efficiencies over traditional Medicare in addressing the same health conditions.”

UnitedHealthcare Ins. Co. v. Becerra, 16 F.4th 867, 873-74 (D.C. Cir. August 13, 2021, reissued

November 1, 2021), cert. denied, 142 S. Ct. 2851 (U.S. June 21, 2022) (No. 21-1140). The risk

adjustment model used for MA payment is not designed to drive clinical behavior to look for

specific conditions or to be the sole purpose for MA organizations or health care providers to

identify and treat conditions that are potential precursors to adverse medical events or

complicating factors in the identification and treatment of other conditions. MA organizations

are required, by their contracts with CMS and section 1852(a) of the Act, to furnish medically

necessary Part A and Part B benefits to their enrollees. Because MA organizations are at

financial risk for the care of their enrollees, changes in the risk adjustment model do not change

the fundamental incentive in a capitated payment system to reduce morbidity and mortality by

identifying and treating early stages of disease. Therefore, we respectfully disagree with

commenters that the changes to the mappings of certain diagnoses or conditions to the HCCs in

the model will disincentive MA organizations from identifying and treating conditions early in

the disease process.

MA organizations submit bids to CMS that request the revenue needed to cover the expected per

beneficiary costs of their enrollee population. Risk adjustment is used to adjust plan bids and

payment based on health status and demographic characteristics such that plans are paid more for

beneficiaries predicted to have higher costs. To accurately predict the likely relative cost of each

beneficiary, it is important to include in the risk adjustment model those diagnoses and

conditions that are reliable predictors of future costs and exclude those that are unreliable

predictors of future costs. For the CY 2024 model, CMS undertook a comprehensive and

thoughtful process, informed by clinical input, to determine the diagnoses and conditions for

inclusion. As a result, the new model will better direct resources to plans with beneficiaries with

higher health care needs.

CMS uses ten longstanding principles, originally introduced in the 1999 Report to Congress and

later expanded and formally established by the 2000 Report to Congress, to group diagnosis

codes into HCCs and select which HCCs are included in the model.

12,13

Regarding the specific

diagnoses and conditions that were included in the updated model, CMS followed these

longstanding principles to group diagnosis codes and build all new HCCs using ICD-10-CM

1999 Report to Congress: Proposed Method of Incorporating Health Status Risk Adjusters into Medicare+Choice Payments

2000 Report to Congress: Diagnostic Cost Group Hierarchical Condition Category Models for Medicare Risk Adjustment

diagnosis codes. ICD-10-CM diagnosis codes have been used by U.S. providers since October

2015. Moving from ICD-9-CM to ICD-10-CM was a major classification change in the U.S.,

both in terms of the volume of diagnosis codes and in the structure and clinical specificity of

codes, as well as changes in clinical concepts for some conditions. Therefore, we conducted

multi-years analyses (2016/2017, 2017/2018, and 2018/2019) to assess the stability of using

ICD-10 diagnosis codes for the model calibration.

Because providers are no longer using ICD-

9, in order to update the underlying data for the model calibration, a reclassification had to occur

to develop HCCs aligned with the ICD-10 classification system. As part of this process, we

determined which diagnoses were the best predictors of prospective costs based on our

longstanding principles for developing risk models.

As a result of this reclassification, the

number of HCCs in the payment model increased from 86 to 115. The updates made to the CMS-

HCC model improves its accuracy by reflecting more recent cost and utilization patterns, and by

including conditions that reliably predict Medicare costs, and thus help ensure that higher

payments are available to plans that serve beneficiaries who are expected to have more costly

health care needs.

Relative factor differences between the current model and the proposed model will occur for

several reasons. Since the model update included updated data years and included an HCC

reclassification, changes in the marginal cost attributable to each HCC is driven by both of these

updates. In addition, the relative factors associated with each condition is relative to the change

in average cost (i.e., the updated denominator). As noted above, the risk model was reclassified

increasing the number of HCCs from 86 in the current model to 115 in the updated model. For

this reason, generally comparing HCC coefficients across both models would not be an

appropriate measurement. For those conditions in which the HCC persisted across both models,

some coefficients have increased more than average and some have increased less than average.

For those conditions whose dollar coefficients have increased more than average, the relative

coefficients have increased (see Multiple Sclerosis example discussed earlier in this section). As

noted in the CY 2024 Advance Notice, changes to the condition categories – including additions,

deletions, and revisions – are based on each condition category’s ability to predict costs for

Medicare Parts A and B benefits. Condition categories that do not predict costs well – for

example, because the coefficient is small, the t-value is low, a small number of beneficiaries

have the certain condition, or the condition does not have well-specified diagnostic coding – are

not included in the model. As previously stated, condition categories and the underlying ICD-10

to HCC mappings are evaluated using the longstanding risk adjustment principles. Empirical

data are taken into account along with clinician information and clinician input when evaluating

HCC for inclusion in the model. Empirical data analysis includes reviewing diagnosis

In Section 4 Ongoing Research of the 2018 & 2021 Reports to Congress CMS discusses the need to evaluate ICD-10 diagnosis

patterns: https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Risk-Adjustors-Items/ReportToCongress

Advance Notice of Methodological Changes for Calendar Year (CY) 2004 Medicare+Choice (M+C) Payment Rates; 2021

Report to Congress on Risk Adjustment in Medicare Advantage (section 2.3 provides further details on the Principles of Risk

Adjustment). See also 2018 and 2021 Reports to Congress

frequencies, predictive power, expenditures associated with the condition, chronicity (duration of

a condition), and prospective cost prediction (analyses of concurrent versus prospective costs

implications).

To obtain clinical input throughout the model development process, CMS consulted with a panel

of internal medicine and general practitioners to provide clinical expertise, as the model covers

all body systems and a broad spectrum of diagnoses. The clinicians are practicing physicians, as

well as professors, and health services researchers. A series of discussions with the clinicians

were held over the course of a year to review and discuss the reclassification. The reclassification

process, including the clinical consultation panel review, is an iterative process. The process

involves empirical data investigation and analysis by healthcare economists, clinical research,

and clinician input and review. Multiple iterations of the model are calibrated to evaluate

diagnosis mappings, HCC reconfigurations, clinical hierarchies, and which HCCs are included in

the model for payment. Clinician input was provided, where applicable, for each model iteration.

Decisions regarding the model classification (e.g., how diagnoses are grouped) were made based

on the level of granularity provided in the ICD-10 classification system, the risk adjustment

model principles, examination of the model performance (i.e., predictive accuracy), and for the

initial analysis to further assess which conditions are more discretionary, a review of HCC

frequencies in MA and FFS.

The clinical panel provided insight and recommendations related to the criteria used to evaluate

and treat conditions (e.g., lab findings, test results, physical exams, medications, medical

interventions, etc.); specificity of diagnoses; potential for variable coding for certain diseases or

disease groups; treatment practices; manifestations of conditions in different populations (e.g.,

aged vs. disabled; non-dual vs. full-benefit; dual vs. partial-benefit dual); disease severity and

implications for medical burden; the need for additional medical specialist intervention (i.e.,

whether or not a condition would require evaluation by a specialty clinician); clinical face

validity (i.e., whether the condition is consistent with the relative costs associated with it) and

interpretation of empirical results; clinical similarities and differences between specific diseases

(e.g., Crohn’s disease and ulcerative colitis); relationship of acute and chronic versions of

diseases (e.g., acute hepatitis C and chronic hepatitis C); and implications for predicted costs.

The HCC classification system is exhaustive (i.e., all diagnosis codes map to an HCC). While all

diagnosis codes map to an HCC, only a smaller subset of HCCs are included in the payment

model. When CMS refers to “payment HCC” we are referring to an HCC that is included in the

model for payment, and vice versa for references to “non-payment HCCs.”

Below we address the specific conditions that were re-mapped or grouped differently in the

updated 2024 risk adjustment model and about which concerns were raised by commenters.

There were HCC reconfigurations and underlying diagnosis changes as described in the 2024

Advance Notice. Below we review these changes:

• Vascular Disease (Diagnoses related to atherosclerosis, aortic, deep vein thrombosis,

embolism, and other circulatory diseases)

o As identified in the 2024 Advance Notice, the Vascular disease group was reconfigured

from the 2020 CMS-HCC model to the 2024 CMS-HCC model. There are 365 ICD-10

payment diagnosis codes in the vascular disease group. The increased granularity under

the ICD-10 classification system permitted better configuration of the vascular HCCs,

and HCCs 107-108 were reconfigured into three new HCCs, 263, 264, and 267. Under

the current 2020 CMS-HCC model, vascular HCC 108 contains a range of ICD-10 codes,

including those describing atherosclerosis of the aorta, renal artery, and extremities,

except with ulceration or gangrene; specified aneurysms; and peripheral vascular disease.

ICD-10 codes atherosclerosis of arteries of extremities unspecified, with exercise or rest

pain, ulceration, or gangrene are mapped to HCCs in the Vascular disease hierarchy

according to their clinical severity. Based on empirical data (e.g., expenditure patterns,

frequency, predictive accuracy) and clinician input, payment HCCs for atherosclerosis of

arteries of extremities in the updated 2024 CMS-HCC model contain primarily codes for

atherosclerosis with rest pain and with ulceration or gangrene, as they typically indicate

more serious cases.

o In consultation with clinicians, codes that indicate more serious cases of vascular disease

were retained in the payment HCCs, and diagnoses for “other” and “unspecified” disease

were mapped to a non-payment HCC. Diagnosis codes for atherosclerosis of arteries of

the extremities with rest pain were moved to the higher-level payment HCC 264.

Diagnoses “with rest pain” indicates compromised vasculature in the leg and that

vascular disease is likely more widespread, so usually requires intervention. Codes for

atherosclerosis of arteries of the extremities, with intermittent claudication were broken

out as new nonpayment HCC 265 (refer to additional information provided in a later

response). A new HCC 267 was created for Deep Vein Thrombosis and Pulmonary

Embolism due to clinical considerations (similar indications for treatment and cost

implications in a prospective model). Of the remaining codes from 2020 CMS-HCC

model HCC 107, only the more severe manifestations that would consistently and

reliably predict costs prospectively (e.g., aortic aneurysm, ruptured and arterial embolism

and thrombosis) are mapped to payment HCC 264. Diagnosis codes (e.g., Atherosclerosis

of the Aorta) for manifestations or stages of diseases that are not consistently reliably

predictive of future years costs (e.g., involve lifestyle changes, or medical monitoring)

were mapped to the lower level Vascular HCC. The new treatment of the vascular

disorders is consistent with the treatment of coronary atherosclerosis in the CMS-HCC

model, in which the diagnosis of coronary atherosclerosis is non-payment, but the more

severe manifestations of unstable angina and acute myocardial infarction are payment

HCCs.

• Metabolic Diseases (Endocrine related diagnoses)

o There are three payment HCCs in the 2020 CMS-HCC model Metabolic disease

group. The 2024 CMS-HCC model Metabolic disease group was expanded to four

HCCs, in the payment model. HCC 23 (Other Significant Endocrine and Metabolic

Disorders) was reconfigured to pull out select conditions with prospective cost

implications. There are 59 ICD-10 payment diagnosis codes in the metabolic disease

group. First, high-cost lysosomal storage disorders with expensive Part B drug

treatments were split out into a new HCC 49 Specified Lysosomal Storage Disorders.

Then metabolic disorders and endocrine disorders were separated based on cost and

clinical considerations into HCC 50 Amyloidosis, Porphyria, and Other Specified

Metabolic Disorders, and HCC 51 Addison’s and Cushing’s Diseases, Acromegaly,

and Other Specified Endocrine Disorders, respectively. Other conditions (e.g.,

aldosteronism) from the 2020 CMS-HCC model HCC 23 that may be more indicative

of lab test results and can be manifestations of other, underlying disorders (e.g., heart

failure), or are primarily treated with lower-cost medications were mapped to a non-

payment HCC based on empirical overprediction (i.e., overprediction of the costs for

beneficiaries with these conditions) and clinical considerations. This is consistent

with the treatment of hypothyroidism and hyperthyroidism, which are not in the 2020

or proposed 2024 CMS-HCC payment models.

• Heart Diseases (Cardiovascular and heart failure diagnoses)

o There are five payment HCCs in the current 2020 CMS-HCC model Heart disease group.

In the reclassified 2024 CMS-HCC model, the Heart disease group was expanded to ten

payment HCCs, due to the split-out of 2020 CMS-HCC model HCC 85 Congestive Heart

Failure into five payment heart failure HCCs (222-226) to account for clinical severity

and cost differences. There is a hierarchy among heart failure HCCs, with new HCC 221

Heart Transplant Status/Complications added to the top of the hierarchy and new split-

out HCC 227 Cardiomyopathy/Myocarditis at the bottom. Multiple different

configurations of heart failure codes were evaluated, and the implemented hierarchy was

finalized based on empirical analyses and consulting clinicians’ input. The HCC structure

captures differences in prospective costs among conditions and has clinical salience.

Principle 10 focused changes are discussed in a later response.

• Blood Disease (Coagulation diagnoses and disorders of immunity)

o There are three payment HCCs in the 2020 CMS-HCC model Blood disease group. In the

2024 CMS-HCC model, the Blood disease group was expanded to seven payment HCCs.

There are 132 ICD-10 payment diagnosis codes in the blood disease group. Specifically,

diagnoses that describe coagulation defects, hemorrhagic conditions, and purpura were

mapped to payment HCC 112 Immune Thrombocytopenia and Specified Coagulation

Defects and Hemorrhagic Conditions or non-payment HCC Thrombocytopenia, Purpura,

Thrombophilia, and Other and Unspecified Hemorrhagic Conditions based on clinical

severity, specificity, and the reliability for predicting prospective costs. There are two

payment HCCs (114-115) in the 2024 CMS-HCC model Blood disease group that

describe immune conditions. HCCs 114 Common Variable and Combined

Immunodeficiencies and 115 Specified Immunodeficiencies and White Blood Cell

Disorders contain immune conditions from the 2020 CMS-HCC model HCC 47. This

split-out allows for isolation of the costlier and clinically severe common variable and

combined immunodeficiencies (in 2024 CMS-HCC model HCC 114) from other

specified immunodeficiencies and white blood cell disorders (in 2024 CMS-HCC model

HCC 115), while unspecified, non-specific, and lower-severity immune disorders were

mapped to a non-payment HCC.

• Amputation

o There is one payment HCC in the 2020 CMS-HCC model Amputation disease group,

which was reconfigured to cover initial complications or ongoing costs of lower limb

amputation in the 2024 CMS-HCC model. There are 44 ICD-10 payment diagnosis codes

in the amputation disease group. Diagnosis codes for the acquired absence codes of toe

and finger were mapped to the non-payment HCC Post-Surgical States/Aftercare/Elective

to more accurately classify them based on similar implications for disease burden (e.g.,

similar clinical implications for care) and cost prediction. This is parallel to the RxHCC

model, where these codes similarly map to nonpayment HCCs.

• Neurological Diseases (Polyneuropathy diagnoses)

o There are eight payment HCCs in the current 2020 CMS-HCC model Neurological

(Neuro) disease group. In the 2024 CMS-HCC model, the Neuro disease group was

expanded to twelve payment HCCs. There are 200 ICD-10 payment diagnosis codes in

the neurological disease group. Codes in the current 2020 CMS-HCC payment HCC 75

Myasthenia Gravis/Myoneural Disorders and Guillain-Barre Syndrome/Inflammatory

and Toxic Neuropathy were reconfigured into 2024 CMS-HCC model HCCs 193-196.

First, the most underpredicted and chronic codes were pulled out into new payment HCC

193 Chronic Inflammatory Demyelinating Polyneuritis and Multifocal Motor

Neuropathy. The acute Guillain-Barre Syndrome became non-payment HCC 194 in the

2024 model; costs associated with this HCC are predicted reasonably well without being

in the payment model, and this diagnosis will transition to HCC 193 if it becomes long-

lasting/chronic. Myasthenia gravis codes were reconfigured into two payment 2024

CMS-HCC model HCCs based on clinical severity and cost differences: HCC 195

Myasthenia Gravis with (Acute) Exacerbation; and HCC 196 Myasthenia Gravis without

(Acute) Exacerbation and Other Myoneural Disorders. In the 2024 CMS-HCC model,

The Rx Hierarchical Condition Categories (RxHCC) model is used in Medicare Part D risk adjustment.

https://www.cms.gov/files/document/2023-announcement.pdf.

nonpayment HCC 203 Polyneuropathy, Mononeuropathy, and Other Neurological

Conditions/Injuries remains parallel to the 2020 CMS-HCC model nonpayment HCC 81,

with the addition of several clinically related codes determined not appropriate for the

payment model on empirical, clinical, or coding specificity grounds (narcolepsy and

cataplexy, drug-induced and other polyneuropathies, neuromyopathy and central nervous

system atrophy).

• Diabetes

o There are three payment HCCs in the 2020 CMS-HCC model Diabetes disease group.

The 2024 CMS-HCC model HCC classification has four payment diabetes HCCs, as

HCC 35 Pancreas Transplant Status was added to the top of the hierarchy. There

continue to be three 2024 CMS-HCC model payment HCCs for diabetes with/without

complication, with minor differences in code content compared to the payment HCCs in

2020 CMS-HCC model. There are 344 ICD-10 payment diagnosis codes in the diabetes

disease group. In consultation with clinicians, the diagnosis codes for diabetes with

unspecified complications and with complications related to glycemic (i.e., blood sugar)

control were moved to the lowest payment HCC, 2024 CMS-HCC model HCC 38. This

was due to empirical data regarding coding frequency, as well as clinical considerations

(high blood sugar is the defining characteristic of diabetes, and thus is inherent to the

disorder rather than a complication). Glycemic complications can be acute, side effects of

drug therapy, and of less serious magnitude than complications related to other disease

groups that remain in the 2024 CMS-HCC model HCC 37 Diabetes with Chronic

Complications, and could also be associated with poorer quality of care (poor glycemic

control). Severe acute complications related to glycemic control (e.g., E11641 Type 2

diabetes mellitus with hypoglycemia with coma) remain in the highest HCC, 2024 CMS-

HCC 36. Some drug induced diabetes codes were re-mapped to a non-payment HCC

because the conditions can be temporary/reversible, variable cost profile, and the removal

allows costs to be attributed to conditions that represent the underlying health status.

• Kidney Disease

o There are five payment HCCs in the current 2020 CMS-HCC model Kidney disease

group. In the 2024 CMS-HCC model, there are four payment Kidney HCCs. There is a

straight hierarchy between the payment HCCs. There are 9 ICD-10 payment diagnosis

codes in the kidney disease group. The kidney hierarchy relies on the chronic kidney

disease (CKD) stages to define the payment HCCs. In the 2024 CMS-HCC model, the

2020 CMS-HCC model HCC 138 was replaced with two new, more granular HCCs –

328 Chronic Kidney Disease, Moderate (Stage 3B) and 329 Chronic Kidney Disease,

Moderate (Stage 3, Except 3B) – due to new FY2021 ICD-10 codes that split the broad

range of Glomerular Filtration Rate (GFRs) in stage 3 into 3a, 3b, or 3 unspecified. There

is a specified (a priori) constraint setting HCCs 328 and 329 coefficients equal, so stage 3

will continue to have a single payment coefficient until there is sufficient data in later

years to distinguish, estimate, and evaluate separate coefficients for HCCs 328 and 329 in

future model updates. The HCC-level split-out is parallel to the Part D V08 RxHCC

reclassification that was implemented for CY 2023, and provides future flexibility to

reconsider the constraint at the point when underlying data that includes the separate

codes for CKD stages 3a and 3b is used for a recalibration. Due to clinical considerations,

the 2020 CMS-HCC model HCCs 134 and 135, renumbered as HCC 324 Dialysis Status

and HCC 325 Acute Kidney Failure in the 2024 CMS-HCC model, were removed from

the payment model and hierarchy. HCC 324 Dialysis Status captures transitory dialysis,

and the removal from payment allows the costs to flow through to actual conditions that

created the need for dialysis, such as congestive heart failure or sepsis. If dialysis is

chronic, the beneficiary will attain ESRD status and be moved to risk scores calculated

using the ESRD model. HCC 325 Acute Kidney Injury (AKI) is also by definition

transitory. Beneficiaries without transitory disease should receive a CKD diagnosis. If

CKD stage 3 or above (moderate or severe) develops, it will be captured in the payment

model through the CKD HCCs.

• Psychiatric Diseases (Diagnoses related to Schizophrenia, Psychosis, Personality Disorders,

Bipolar Disorders and Major Depression)

o There are four payment HCCs in the 2020 CMS-HCC model Psychiatric disease group.

The 2024 CMS-HCC model Psychiatric disease group was expanded to five payment

HCCs, which remain in a straight hierarchy. There are 438 ICD-10 payment diagnosis

codes in the psychiatric disease group. The top HCC – renumbered from 2020 CMS-HCC

model HCC 57 to 2024 CMS-HCC model HCC 151 Schizophrenia – remains unchanged.

The 2020 CMS-HCC model HCC 58 (Reactive and Unspecified Psychosis) was

reconfigured as 2024 CMS-HCC model HCC 152 Psychosis, Except Schizophrenia.

Several conditions – delusional disorders, manic and depressive (bipolar) disorders,

severe, with psychotic behavior, and major depressive disorder, severe, with psychotic

features – were moved out of the HCC 59 in the 2020 CMS-HCC (current) model into

HCC 152 in the 2024 CMS-HCC model because costs for beneficiaries with these

conditions were underpredicted in the 2020 CMS-HCC model depression HCC 59 and

are clinically consistent with non-schizophrenic psychosis. The 2024 CMS-HCC model

HCC 153 Personality Disorders; Anorexia/Bulimia Nervosa, was reconfigured from

2020 CMS-HCC model HCC 60. Personality disorders were moved up above non-

psychotic depression/bipolar disorders in the reclassified Psychiatric hierarchy because of

their higher associated prospective costs. Consulting clinicians supported this

reorganization as it is reflective of care practices associated with the conditions.

Diagnosis codes for anorexia/bulimia nervosa were moved into 2024 CMS-HCC model

HCC 153 (a payment HCC) based on empirical underprediction in a non-payment 2020

CMS-HCC and consulting clinicians’ input that these conditions have prospective cost

implications since they require specialized treatment and frequent institutionalization.

The subsequent two payment 2024 CMS-HCC model HCCs (154-155) relate to

depression and bipolar disorders. The 2020 CMS-HCC model HCC 59 Major

Depressive, Bipolar, and Paranoid Disorders was reconfigured. In addition to the codes

specifying psychosis that were moved up to 2024 CMS-HCC model HCC 152, only

select other codes from the 2020 CMS-HCC model HCC 59 were retained as codes used

in risk adjustment payment, in the 2024 CMS-HCC model HCC 154 Bipolar Disorders

without Psychosis and HCC 155 Major Depression, Moderate or Severe, without

Psychosis. Bipolar disorder was split out as HCC 154 to reflect the greater clinical

severity compared to depression, which it hierarchically excludes. As discussed in the

“Clinical Update using ICD-10 Diagnosis Code” section of the 2024 Advance Notice, the

reclassification was guided by the longstanding 10 risk adjustment principles, including

that diagnoses in the payment model be clinically meaningful (well specified) and should

reliably predict costs in a future year. As such, diagnoses for mild, unspecified,

remission, subsequent encounter, and sequela codes were re-mapped to appropriate non-

payment HCCs. For depression, the 2024 CMS-HCC payment model is limited to ICD-

10 codes specifying moderate or severe major depression. Remaining diagnosis codes

(includes depression codes for mild, unspecified, or in remission, as well as codes for

bipolar disorders in remission) from the 2020 CMS-HCC model HCC 59 were remapped

to the nonpayment depression HCC. These codes represent less-severe conditions with

lower cost implications, and clinicians expected depression would be coded as moderate

or severe if the condition was sufficiently serious. Similarly, codes specifying in

remission were removed as beneficiaries experiencing relapse would likely be coded with

a diagnosis reflecting an active disorder. Diagnosis codes from the 2020 CMS-HCC

model HCC 59 that specify subsequent encounter for or sequela of attempted

suicide/intentional self-harm, were removed from payment, consistent with the removal

of most subsequent encounter and sequelae codes throughout the reclassified 2024

CMS-HCC model. This accounted for the majority of codes from 2020 CMS-HCC model

HCC 59 that were removed from payment in the 2024 CMS-HCC model (>95%).These

codes were remapped to the nonpayment External Causes of Morbidity and Injury disease

groups, as subsequent encounter codes require less intensive follow-up, and sequela

codes are for unspecified late effects of a condition, which are variable in clinical burden

and can be acute, and do not reliably or consistently predict meaningful prospective costs.

(The specific condition that is a sequela should be coded separately, and could enter the

payment model on that basis.) As identified in the underlying ICD-10 diagnosis to HCC

mappings CMS released with the 2024 Advance Notice, the initial encounter diagnosis

codes for attempted suicide/intentional self-harm conditions that are being actively

treated and predict ongoing and prospective costs still map to payment HCCs.

• Musculoskeletal Diseases (Diagnoses related to rheumatoid arthritis and connective tissues).

o There are two payment HCCs in the current 2020 CMS-HCC model Musculoskeletal

(MSK) disease group. There are three payment HCCs in the 2024 CMS-HCC model

MSK disease group. There are 1,187 ICD-10 payment diagnosis codes in the

musculoskeletal disease group. The 2020 CMS-HCC model HCC 40 was split into 2024

CMS-HCC model HCC 93 Rheumatoid Arthritis and Other Specified Inflammatory

Rheumatic Disorders and HCC 94 Systemic Lupus Erythematosus and Other Specified

Systemic Connective Tissue Disorders according to the underlying expenditures

associated with these conditions. In consultation with clinicians the conditions were split

rather than combined as they vary in chronicity, severity, and progression. HCC 93

hierarchically excludes HCC 94 so a beneficiary may receive at most one of these two

HCCs. Underpredicted codes (i.e., diagnoses where the model underpredicts costs for

beneficiaries with the condition) were moved from the 2020 CMS-HCC model

nonpayment Neuro (inclusion body myositis) and MSK HCCs (sarcoid arthropathy and

sarcoid myositis) to the payment HCC 93 in the 2024 CMS-HCC model to improve

prediction and enhance clinical coherence (i.e., grouping diagnoses that are clinically

similar). Clinically relevant, cost-similar codes were moved from a payment 2020 CMS-

HCC model Vascular HCC into 2024 CMS-HCC model HCC 94 (other and unspecified

necrotizing vasculopathies). Conditions (e.g., polymyalgia rheumatic, sacroiliitis, and

Sicca (Sjogren’s) syndrome) that were overpredicted (i.e., the mode overpredicted costs

for beneficiaries with these conditions) were moved to a non-payment HCC based on

input from consulting clinicians. For example, Sicca syndrome and sacroiliitis are

localized, whereas other conditions from the 2020 CMS-HCC model HCC 40 are

systemic, and treatment for these conditions is symptom-driven (e.g., eye drops or

sucking candy for Sicca) and unlikely to be associated with prospective costs.

Polymyalgia rheumatica can be clinically variable, is often treated by medications, and is

likely to be picked up by other diagnosis codes in severe cases.

In summary, the CMS-HCC reclassification involved revising condition categories – including

adding, deleting, and reconfiguring categories and clinical hierarchies, and freshly considering

which categories are included in the payment model. The goal was to improve predictive ability,

to better account for current disease patterns, treatment methods and costs, and diagnosis and

coding practices. The resulting model classifies the ~74,000 ICD-10-CM codes into 266 CMS-

HCCs, 115 of which are included in the 2024 CMS-HCC payment model. This increase in

condition categories from the current 2020 CMS-HCC model (204 CMS-HCCs; 86 in payment)

is due to the greater level of detail in ICD-10-CM diagnosis codes, allowing for the development

of HCCs with increased clinical specificity and validity that better capture clinical and cost

differences between conditions. In aggregate, the 2024 CMS-HCC model contains approximately

20 percent fewer ICD-10-CM codes than the 2020 CMS-HCC model. This resulted from the

removal of diagnoses in accordance with CMS’ risk adjustment principles, evaluated based on

(1) empirical data including frequency, sample size, associated expenditures (e.g., overpredicted

under the current model HCC); (2) clinical specificity and salience; (3) reliability to predict

prospective costs (including conditions that represent side effects of medical or drug treatment

rather than underlying health status risk); and (4) variable diagnosis or reporting (based on

empirical evidence or clinical input). Table III-3 below provides additional information on the

underlying diagnosis code counts for the current 2020 CMS-HCC model and the 2024 CMS-

HCC model.

Table III-3 Summary Statistics for the 2020 CMS-HCC and 2024 CMS-HCC Classifications

2020 CMS-HCC

Model

2024 CMS-HCC

Model

FY22/23 ICD-10 codes - total

73,926*

FY22/23 ICD-10 codes mapped to payment HCCs

9,797 (13.3%)

7,770 (10.5%)

FY22/23 ICD-10 codes mapped to non-payment HCCs

64,129 (86.7%)

66,156 (89.5%)

Added

209

No longer mapped in the 2024 CMS-HCC Model

2,236

No longer mapped – ICD-10 clinical updates

2,161 (96.6%)

No longer mapped – Principle-10 focused updates

75 (3.4%)

HCCs - total

204

266

HCCs – payment

86 (42.2%)

115 (43.2%)

HCCs – non-payment

118 (57.8%)

151 (56.8%)

* The total number of ICD-10 diagnosis codes varies by fiscal year.

Comment: A few commenters stated concerns that the Fact Sheet released with the Advance

Notice did not discuss the methodology, assumptions, and data used for developing the MA risk

score trend. A commenter believed that not providing this additional context may lead to

confusion, misinterpretation, and possibly false conclusion about the impact of the risk score

trend on MA payments. Commenters requested that CMS include the methodological details

behind the MA risk score trend.

Response: Each year, CMS releases an associated Fact Sheet that shows the year-to-year

percentage change in payment associated with the proposed (in the Advance Notice) or finalized

(in the Rate Announcement) policies. The Fact Sheet shows the overall average impact on MA

revenue, as well as the average impact of each individual update or policy proposal. As part of

the impacts released in the Fact Sheet, CMS also estimates the average growth of MA risk scores

in the payment year, known as the MA risk score trend. The MA risk score trend is the average

increase in risk scores, not accounting for normalization and the MA coding pattern adjustment

(which are included separately). The MA risk score trend is included in the Fact Sheet because it

has direct bearing on MA payments and the MA revenue picture would be incomplete without it.

CMS calculates the MA risk score trend by calculating MA risk scores over three prior years,

then calculating the average annual change in risk scores across those three years. All three years

of risk scores are calculated using the risk adjustment model(s) to be used in the upcoming

payment year. This average annual change is the MA risk score trend provided in the Advance

Notice and Rate Announcement Fact Sheet. The trend is an industry average and individual

plans’ experience will vary.

Note that the MA risk score trend has a separate impact from the impact of the risk adjustment

model, which is represented in a separate row of the Fact Sheet, and is based on risk score

changes where the underlying data is held steady. Specifically, to measure the risk adjustment

model impact CMS uses the same diagnostic and demographic information run through the

current model and the model(s) for the upcoming payment years (e.g., 2020 diagnoses were used

to calculate 2021 risk scores under each model to calculate the risk adjustment model impact in

the Fact Sheet). The difference between the current model and payment year model(s) risk

scores, accounting for differences in normalization, is represented in the risk adjustment model

impact and normalization row.

By including both the risk adjustment model/normalization impact and the MA risk score trend

in the Fact Sheet, the resulting impact is effectively estimating a year-over-year payment impact

if diagnostic and demographic data are held steady, then further accounting for growth in risk

scores in the payment year based on historical experience. Therefore, it is imperative to consider

the MA risk score trend in concert with the impact of risk adjustment policy proposals to

accurately predict payment impacts in the following year.

It is important to note that every model has its own risk score trend. The current model used for

payment in CY 2023 (the 2020 risk adjustment model) has a risk score trend of 5%. Because we

are blending risk scores from the updated 2024 model with risk scores from the current 2020

model, the effective MA risk score trend of 4.44% is in the Fact Sheet for the CY 2024 Rate

Announcement. This means that though we anticipate an increase in risk scores based on the

diagnoses submitted by plans, the impact on payment is anticipated to be less in the updated

model.

Comment: Many commenters expressed concern that the updated model may have a negative

impact on certain beneficiary populations, locations, and plan types. Multiple commenters

believed that the diabetes changes and model changes where diagnoses are no longer included

for payment will have a negative impact on dually eligible beneficiaries and vulnerable

populations (e.g., minority beneficiaries and those under the federal poverty level), or

beneficiaries in urban or rural areas.

Multiple commenters expressed concerns about the model’s impacts on plans with high

enrollment of dually eligible beneficiaries, and high-risk, chronically complex vulnerable

populations (e.g., Special Needs Plans that serve dually eligible beneficiaries (D-SNPs) or

beneficiaries with certain chronic diseases (C-SNPs). Multiple commenters stated that minorities

and people with low incomes make up a larger share of MA enrollees than they do in FFS, and

that the proposed model will more negatively affect them. Commenters expressed concern about

the implication for benefits that SNP beneficiaries have, such as greater access to supplemental

benefits, and stated that, according to their research, the proposed model will have a more

negative impact on D-SNPs and C-SNPs, therefore negatively impacting the beneficiaries’

access to these benefits. Commenters also expressed concern that the negative impact of the

proposed model may disincentivize MA plans from offering C-SNPs. A few commenters

believed the updated model will also negatively impact employer-group waiver plans (EGWPs)

with a commenter noting the inability for EGWPs to adjust bids in response to potentially lower

risk scores.

Response: The updates improve the accuracy of the risk adjustment model and help ensure that

higher payments are available to plans that serve beneficiaries with more costly health care needs

(refer to the predictive accuracy information provided above). Additionally, the updates do not

change features in the CMS-HCC risk adjustment model, first implemented in 2017, that ensure

dually eligible beneficiaries have unique adjustments for every health condition based on their

dual-eligibility status that result in higher payments for those conditions than non-duals. The

updates also do not alter changes first implemented in 2020 that ensure that plans receive an

additional increase in payment based on the number of conditions the beneficiary has.

As previously discussed in this Rate Announcement, conditions in the model are used as

predictors of relative costs, not as direct reimbursement for the treatment of each condition. Plan

bids project the average revenue needed to cover all Part A and B benefits, and the risk score is

used to assess the relative cost of a plan’s enrollee population. Further, it is the total risk score

that predicts the relative cost of a beneficiary, and each factor predicts part of the costs;

therefore, each relative factor cannot be assessed in isolation. If a specific HCC (or diagnosis

code mapped to a specific HCC) is no longer included in the payment model, coefficients of

other HCCs and demographic factors will be increased such that the model continues to predict

the overall total expenditures (please see above for a discussion of the model’s predictive

accuracy over subgroups of beneficiaries by level of risk). Because the updated model improves

upon the current model by incorporating recent costs and utilization patterns and is developed

using ICD-10 codes, and because the model ensures that plans that enroll higher need

beneficiaries receive higher payments, we do not agree that the updated model will negatively

affect beneficiary costs or supplemental benefits, and care delivery (see also response to

comments above).

There will be variation in the impact on risk scores depending on each beneficiary’s clinical mix.

All of the model updates (i.e., underlying data updates, denominator update, and ICD-10

reclassification) contribute to changes in the relative costs of conditions compared to the 2020

CMS- HCC model currently used, and therefore changes to the resulting risk scores. Beneficiary

risk scores or plan average risk scores may change depending on an individual beneficiary’s

combination of diagnoses or the clinical profile of a plan’s enrollee population. CMS has

observed that, on average, predicted risk for dually-eligible populations and those in special

needs plans are higher than non-dually-eligible enrolled beneficiaries. On average, under the new

CY 2024 risk adjustment model, risk scores for special needs plans (SNPs) are 50% higher than

non-SNPs. More specifically, risk scores for C-SNPs and D-SNPs are, respectively, 54% and

47% higher than non-SNP beneficiaries. Risk scores, under the 2024 updated model, for dually-

eligible beneficiaries in the community are on average 49% higher compared to non-dually

eligible beneficiaries in the community. While risk scores change in reflection of the underlying

changes in the relative factors associated with each condition, on average, the model continues to

predict higher risk for dually-eligible and SNP enrolled beneficiaries.

In addition, as we did at an industry level, CMS calculated the MA risk score trend for dually-

eligible beneficiaries (i.e., full-dual and partial-dual beneficiaries residing in the community) and

found that the MA risk score trend for these dually eligible beneficiaries is 4.67 percentage

points higher than it is for non-dually eligible beneficiaries. As explained in detail in a later

response, the risk score trend represents the average growth in MA risk scores for the payment

year. As discussed in more detail in a later response in this section, it is imperative to consider

the MA risk score trend in concert with the impact of the updated risk adjustment model (as well

as other changes to payment factors) to accurately predict payment impacts in the following year.

On average, this growth in MA risk scores will more than offset the impact of the new risk

adjustment model and normalization for dually-eligible beneficiaries.

Comment: Commenters stated concern with the CMS review of conditions that focused on

Principle 10 – Discretionary diagnostic categories should be excluded from payment models.

Some commenters noted that removal or constraining of conditions described under Principle 10

may exacerbate health inequities, constrain investments in prevention of severe conditions, and

limit the prevention of subsequent medical events.

Some commenters stated they are supportive of CMS’ decision to update the risk adjustment

model based on more current underlying data, updating the denominator, and some clinical

revisions based on ICD-10, but were specifically opposed to the clinical revisions made under

Principle 10.

Commenters believed that CMS has not released adequate data underlying its determination to

constrain the coefficients for the diabetes and congestive heart failure HCCs. Multiple

commenters stated that CMS does not explain why imposing constraints on these conditions

helps the predictive value of the model. Some commenters believed that the constraining of the

coefficients for the diabetes HCCs in the updated model would delay diagnosis for diabetics and

lead to serious complications by lowering early detection.

Multiple commenters suggested that the removal of the conditions angina pectoris, protein-

calorie malnutrition, and atherosclerosis with intermittent claudication will also affect the ability

to conduct further clinical investigation and intervention. Some of the commenters believed that

the Principle 10-based changes are not clinically appropriate and do not accurately reflect

beneficiary acuity or cost of care. A commenter stated beneficiaries with angina pectoris incur

more than twice the expenditures of a typical Medicare beneficiary and four times the hospital

costs. A few commenters, citing their clinical experience, oppose removing angina pectoris and

atherosclerosis of lower extremities with intermittent claudication from the updated model

because these conditions can help with early identification and prevention of more serious

conditions such as fatal cardiovascular disease. A few commenters expressed concern that

removing protein-calorie malnutrition from the model, which can indicate frailty, will also

adversely affect institutional level of care of enrollees and beneficiaries with HIV/AIDS.

Multiple commenters cited internal analysis stating vulnerable populations and minority

communities have higher prevalence rates of conditions identified in Principle 10.

Response: CMS appreciates the feedback and concerns of the commenters.

As described in the 2024 Advance Notice, for conditions in the model where coding in MA was

highest relative to FFS, CMS reviewed these conditions with our clinical experts for evaluation

against the model principles because we believe that this coding differential indicates conditions

where there may be discretionary coding variation. The review is consistent with our evaluation

of condition categories, and the underlying ICD-10 diagnosis code to HCC mappings, against

risk adjustment model development Principle 10. In consultation with clinician input regarding

appropriate classification of identified conditions, this reclassification involved moving some

discretionary diagnosis codes from condition categories included in the CMS-HCC model to

condition categories not included in the model, removing from the model several condition

categories that do not accurately predict the projected cost of a beneficiary, and constraining

HCCs to be equal to each other so that they carry the same weight in the risk score. These

updates serve to maintain the integrity of the condition categories in the model.

CMS approached the inherent tradeoffs involved in designing a classification system using

empirical evidence on frequencies and predictive power, clinical input on relatedness, specificity,

and severity of diagnoses, and professional judgment on incentives and diagnostic patterns

relative to the classification system. Plans with an atypical distribution in their patient population

may experience varying risk scores impacts; however, the actual change in risk score of any

beneficiary with one or more of the conditions described below will depend on the totality of

their risk profile, including their demographic factors and other diagnoses.

• HCC constraints (i.e., hold the coefficients of the HCCs equal to each other such that each

HCC carries the same weight):

o Constrained all Diabetes HCCs (HCC 36, 37, and 38). In the current 2020 CMS-HCC

model the top two diabetes HCCs, 17 and 18, are constrained. When developing the 2024

CMS-HCC model, empirical data showed that the 2020 CMS-HCC model HCC 18 had a

substantially higher prevalence in MA than FFS and HCC 19 had a lower prevalence; the

shift of beneficiaries to a higher HCC in the hierarchy led to further evaluation of the

HCCs. Under the clinical criteria used to code HCC 18, beneficiaries with a range of

clinical burden and medical expenditures may be captured in the diabetes with chronic

complications HCC resulting in a diminished capacity for the model to differentiate

between disease severity. For example, there may be instances where a beneficiary is

coded with the more severe manifestation of diabetes based on a laboratory finding (e.g.,

automated urinalysis test that finds protein in the urine) even if the finding is not clinically

significant and has no implications for medical treatment. Thus, the constraint was

expanded to all three diabetes payment HCCs in the 2024 CMS-HCC model. Constraining

the Diabetes HCCs allows expenditures to be attributed to well defined and significant

complications of diabetes captured in other payment HCCs (e.g., chronic kidney disease

and diabetic eye disease).

o Constrained Congestive Heart Failure HCCs (HCCs 224, 225, and 226). In the 2020

CMS-HCC model, all Congestive Heart Failure diagnoses are effectively constrained

because they are included in a single Congestive Heart Failure HCC. In the 2024 CMS-

HCC model in consultation with clinicians the single Congestive Heart Failure HCC was

split into five HCCs to better capture the range of heart failure severity. The most severe

manifestation is left unconstrained, while the less severe split outs of the HCC are

constrained, effectively holding them together as they were under the 2020 CMS-HCC

model while CMS continues to observe clinical variation, coding, and cost patterns for

these HCCs. The newly split out HCC 227 Cardiomyopathy/Myocarditis is not

constrained, but is hierarchically excluded by the heart transplant/failure HCCs 221-226.

• HCC removals:

o HCC 47 Protein-Calorie Malnutrition. The current 2020 CMS-HCC model includes a

single marker (HCC 21) for Protein-Calorie Malnutrition. Empirical data shows that MA

reports severe malnutrition at a lower rate than FFS, moderate malnutrition at a slightly

higher rate, and mild and unspecified malnutrition and cachexia at a much higher rate.

Clinically, the more severe manifestations of protein calorie malnutrition can be useful as

a risk marker for frailty, severe illness, and late- or end-stage disease, and it is likely

correlated with poverty/disadvantaged populations. CMS explored creating two HCCs to

distinguish between severe malnutrition and diagnosis codes for mild, unspecified

malnutrition, and cachexia. However, empirical data show that coding for malnutrition is

variable. Removal of the HCC allows expenditures to flow to the well-specified

underlying conditions (e.g., cancer and HIV).

o HCC 230 Angina Pectoris. In the 2020 CMS-HCC model, HCC 88 Angina Pectoris is

one of five payment HCCs in the Heart disease group. It is at the bottom of a payment

hierarchy with HCC 87 Unstable Angina and Other Acute Ischemic Heart Disease and

HCC 86 Acute Myocardial Infarction. In the 2024 CMS-HCC model the Unstable

Angina and Other Acute Ischemic Heart Disease HCC is in the payment model.

Clinicians and empirical data support that unstable angina (HCC 87) has specific

diagnostic criteria, whereas angina pectoris has diagnostic criteria that can result in

coding in situations with little or no significance (e.g., patients exhibiting only chest pain

may receive this diagnosis) and diagnosis and coding may be variable. Therefore,

unstable angina is maintained in the 2024 payment model, while the more variable

manifestation is now mapped to a non-payment HCC.

• HCC 265 Atherosclerosis of Arteries of the Extremities, with Intermittent Claudication.

The Vascular HCC group was restructured (as described above). In consultation with

clinicians, payment codes are focused on peripheral arterial atherosclerosis with

ulceration, gangrene, or rest pain, as these codes indicate more severe manifestations. The

clinical characteristics of intermittent claudication are variable, with some patients

unaware they have the condition in the absence of screening. Therefore, the variation in

coding and clinical implications of this HCC suggest it is not a reliable predictor of

prospective medical expenditures.

Comment: A few commenters stated that CMS is inappropriately reliant on Principle 10 in

adopting changes to make the model more resistant to variable coding. The commenters

requested that CMS provide more transparency on the clinically specific areas impacted by the

Principle 10 updates. A few commenters stated that CMS did not indicate the extent to which the

agency weighted the tenth principle against the other nine. Multiple commenters stated that CMS

did not provide transparency explaining the tradeoffs and inter-relatedness between Principle 10

and other principles. Specifically:

• A few commenters stated that removing diagnosis codes from the model reduces predictive

power and violates Principle 2 (Diagnostic categories should predict medical expenditures).

• One of the commenters believed CMS constraining diabetes and other conditions goes

against the intent of Principle 4 (In creating an individual’s clinical profile, hierarchies

should be used to characterize the person’s illness level within each disease process, while

the effects of unrelated disease processes accumulate).

• A commenter stated CMS’ updated model violates Principle 5 (The diagnostic classification

should encourage specific coding) by paying unspecified diagnoses at the same rate as

conditions with severe acute complications, which does not encourage specific coding.

• A commenter stated the Advance Notice does not fully articulate the various approaches

considered by CMS in shifting ICD codes to different clinical categories in violation of

Principle 9 (The diagnostic classification should assign all ICD-9-CM and ICD-10-CM

codes (exhaustive classification)).

Response: CMS appreciates the feedback and concerns of the commenters.

In rebuilding the diagnostic classification for the 2024 CMS-HCC risk adjustment model,

Principles 7 (monotonicity), 8 (transitivity), and 9 (exhaustive classification) were followed. For

example, if the expenditure weights for the model did not originally satisfy monotonicity,

constraints were imposed to create a model that did. Empirical data and clinical assessment were

used to make tradeoffs among other principles. For example, clinical meaningfulness (Principle

1) is often best served by creating a very large number of detailed clinical groupings. But a large

number of groupings must be balanced with adequate sample sizes for each category (Principle

3). Another tradeoff is encouraging specific coding (Principle 5) versus predictive power

(Principle 2). In ICD-10 there are a number of nonspecific codes. If all of the HCCs these codes

are mapped to are included for payment, the relative cost associated with them would be

marginal and less meaningful. The inclusion of HCCs and associated diagnoses must be balanced

with maintaining or improving the predictive power of the model. CMS approached the inherent

tradeoffs involved in designing a classification system using empirical evidence on the frequency

of medical conditions in MA compared to FFS, predictive power, clinical assessment on

relatedness, specificity, and severity of diagnoses, and professional judgment on incentives and

likely provider responses to the classification system. In developing the 2024 CMS-HCC model,

CMS balanced these competing goals to achieve a risk adjustment model that would result in

payments to MA plans that fairly compensate for the higher costs of sicker enrollees while

paying less for healthier enrollees at an aggregate level.

Comment: A commenter believed CMS may have inadvertently introduced bias in the updated

model by only focusing on the coding frequency of diagnoses in MA compared to FFS for the

determination to exclude or include particular diagnoses based on Principle 10. The commenter

recommended CMS either also consider comorbidity in the MA population compared to FFS, not

consider the relative frequency of coding in MA compared to FFS, or find ways to include

incentives for providers not currently in risk arrangements to screen and treat at the same rates as

seen in MA.

Response: We respectfully disagree with the commenter that Principle 10-focused clinical

updates made to the updated risk adjustment model only focused on reviewing the coding

frequency of diagnoses in MA compared to FFS. As explained in the 2024 Advance Notice (at

47), “[f]or conditions in the model where coding in MA was highest relative to FFS, CMS

reviewed these conditions with our clinical experts for evaluation against the model principles

because we believe that this coding differential indicates conditions where there may be

discretionary coding variation.” Also, as stated above, CMS approached the inherent tradeoffs

involved in designing a classification system using empirical evidence on frequencies, predictive

power, clinical input on relatedness, specificity, and severity of diagnoses, and professional

judgment on incentives and diagnostic patterns relative to the classification system.

Comment: Some commenters stated their belief that CMS is incorrect in our analysis of the

impact of the model and that their analysis indicates the model to have a more negative impact

ranging between -3.4 to -3.7% over the whole industry.

Response: We respectfully disagree with commenters. The industry average calculated by CMS

takes into account the change in risk scores compared to the current risk adjustment model and

the updated risk adjustment model, and necessarily takes into account normalization. As with

every update of the risk adjustment model, the impact on each plan can vary, depending on the

clinical profiles of their enrollees.

Comment: A few commenters stated their concern about the impact of the proposed model on

Puerto Rico. All of these commenters stated that the proposed model will have the largest

negative effect on Puerto Rico due to the territory’s very high MA penetration, poverty levels,

and higher than national average prevalence rates for diabetes, mental health disorders, and

congestive heart failure. A commenter inquired if CMS evaluated how the impact correlates to

certain metrics like geography and/or evaluate the impact based on the ADI.

Response: The CMS-HCC model is a national model, including large subgroup segments that

capture national variation in costs between the segmented populations. The goal of risk adjusted

payments is to pay accurately using the appropriate relative risk for a beneficiary. There will be

variation in the impact on risk scores depending on each beneficiary’s clinical mix. All of the

model updates (i.e., underlying data updates, denominator update, and ICD-10 reclassification)

contribute to changes in the relative costs of conditions, and therefore changes to the resulting

risk scores. Beneficiary risk scores or plan average risk scores may change depending individual

beneficiary’s combination of diagnoses or the clinical profile of a plan’s enrollee population.

As discussed above, the updated model improves predictive accuracy and helps ensure that

higher payments are available to plans that serve beneficiaries with greater health care needs

(refer to the predictive accuracy information provided above). Conditions in the model are used

as predictors of relative costs, not as direct reimbursement for each condition. As intended, the

model ensures that plans that enroll higher need beneficiaries receive higher payments, and,

therefore, we do not agree that the updated model will disproportionately negatively affect

beneficiaries depending on their region.

We understand that geographically Puerto Rico has a high percentage of beneficiaries with risk

scores calculated using the full benefit dual segment. A previously stated, CMS has observed

that, on average, predicted risk for dually-eligible populations are higher than non-dually-eligible

enrolled beneficiaries. As discussed above, the risk score trend for dually-eligible beneficiaries is

higher than the trend for non-dually eligible beneficiaries. As noted in another response, the MA

risk score trend (i.e., average growth in risk scores for the payment year) for dually eligible

beneficiaries is 4.67 percentage points higher than it is for non-dually eligible beneficiaries and

the risk score trend difference for Puerto Rico is even greater. As discussed in more detail in a

later response in this section, it is imperative to consider the MA risk score trend in concert with

the impact of risk adjustment policy proposals to accurately predict payment impacts in the

following year. When the risk adjustment model/normalization and the MA risk score trend are

taken into account, dually-eligible risk scores are less impacted by the risk adjustment changes.

When considering payment, the full scope of contributing factors must be considered. MA plans

submit bids to CMS that request the total revenue needed to cover the expected per beneficiary

costs of their enrollee population. The purpose of the model is to calculate risk scores (that are

used in calculating payments made to plans) to take into account differences in expected costs

for their enrollees and to increase or lower payment based on the relative expected costs. Risk

adjustment is used to adjust plan bids and calculate payments based on health status and

demographic characteristics such that plans are paid more for beneficiaries predicted to have

higher costs due to increased risk. Further, individual coefficients do not represent complete

costs for expected expenditures related to a condition, but only the average increase in the overall

predicted costs for a beneficiary with that condition relative to other conditions used for payment

in the model. Rather, the total relative cost of a beneficiary, or a group of beneficiaries, is

represented by the total risk score.

Comment: Commenters stated their concern that the proposed model will negatively impact

providers engaged in value-based payment models. A commenter noted that many contracts

between plans and their provider partners are based on existing risk models. The commenter

noted the burden that exists by needing to revise these contracts. Some commenters believed the

proposed model may result in physician organizations leaving MA for FFS. A few commenters

noted that the proposed model’s impact on the MA industry will negatively impact what they

believe is a national provider shortage.

Response: CMS thanks the commenters for expressing their concerns. Per section

1854(a)(6)(B)(iii) of the Act, CMS is prohibited from interfering in payment arrangements

between MA organizations and providers with which they contract by requiring specific price

structures for payment. The purpose of the risk adjustment model is to predict the overall relative

expected expenditures for beneficiaries for purposes of paying MA organizations accurately and

fairly for the relative expected costs for the enrollees in their plans. MA organizations in turn

develop provider networks and negotiate payment arrangements with participating providers for

the delivery of covered services to enrollees.

MA organizations are contractually required to cover all Medicare Part A and Part B services

(subject to limited exclusions), maintain adequate networks, and provide quality care. They are

responsible for determining their own revenue needs to cover these services. An updated risk

adjustment model is intended to more appropriately pay plans that are enrolling a sicker

population. In paying plans a capitated payment, CMS contracts with MA organizations for them

to provide coverage of – by furnishing, arranging for, or making payment for – these benefits.

The nature of the MA program, by using capitated payments, allowing MA plans to use a portion

of savings when they bid below the benchmark to furnish additional benefits, and transferring

full financial risk to MA plans, incentivizes MA organizations to develop cost efficiencies in

care provision. Reducing morbidity and mortality by catching early stages of disease in an

inherent expectation of a capitated managed care system.

Section 1852 of the Act requires MA plans to cover Medicare Part A and B benefits (subject to

limited exclusions) for their enrollees and that when the MA plan uses a network of providers

and limits coverage to those providers, the MA plan must ensure that covered benefits are

available and accessible to enrollees. These changes in the CMS-HCC risk model used for risk

adjusting payments to MA plans do not limit or change these requirements related to coverage.

We expect that MA organizations will renegotiate or revise the payment arrangements they have

with their contracted providers as necessary to ensure that the MA plan continues to make

benefits available and accessible for enrollees.

The risk adjustment model is not intended to incentivize (or disincentivize) any particular care

modality. This is illustrated by, for example, not weighting diagnoses by site of care. Another

example is allowing costs to flow to demographic variables, which allows some portion of plan

payments to be paid regardless of disease state and thereby provide funds for a wide range of

prevention and intervention approaches, as well as to cover treatment of acute and lower-severity

chronic conditions not included in the risk-adjustment model. Finally, and to reiterate, by using

more recent data in calibrating the model, coefficients are recalculated, and conditions that might

be relatively more costly than they were before will result in higher risk scores for beneficiaries

with such conditions.

Comment: A few commenters stated their view that it is incorrect to determine that coding

differences between MA and FFS are driven by inappropriate coding.

Response: Discretionary coding involves conditions that are diagnosed and coded inconsistently

across providers and plans. Including conditions in the model where there is variation in their

coding can lead to distortion of the marginal costs estimated by the model, reducing the ability of

the HCCs in the model to predict stable costs and accurately predict those costs in alignment

with the severity of the condition. Diagnosis codes that are subject to discretion are not

necessarily or by definition an indication of inappropriate coding.

Comment: Some commenters recommended different methods for updating the CMS-HCC risk

adjustment model, including making updates on a more regular basis, adding drug utilization to

the model, using more sources of data (e.g., home health and skilled nursing facilities), using a

concurrent model, and calibrating a model using MA encounter data. A number of commenters

recommended social drivers be incorporated into the risk adjustment model. Some commenters

offered recommendations on ways to improve the risk adjustment model by incorporating

additional factors that could improve prediction of the relative costs of MA enrollees by

accounting for social risks. Some commenters recommended CMS integrate all Z-codes, Z55

through Z65, into the HCC risk adjustment model. Other examples of recommendations for

future risk adjustment models include:

• Aligning coding expectations between MA and FFS. This includes the FFS program

rejecting claims until an appropriate diagnosis code is included;

• Adding a frailty measure to the risk adjustment model; and

• Applying a single coefficient to conditions with wide ranges of complexity.

Response: CMS thanks commenters for their suggestions and feedback on updating the CMS-

HCC risk adjustment model. CMS is committed to ensuring the risk adjustment model continues

to perform well in predicting relative risk.

Comment: Some commenters questioned if the updated risk adjustment model will affect lines of

business outside of Medicare Advantage such as the ACO Reach and Medicare Shared Savings

Program.

Response: The Medicare Shared Savings Program and certain CMMI models have incorporated

the CMS-HCC risk adjustment model into their financial methodologies. CMS is considering the

implications of these changes to the CMS-HCC risk adjustment model for these initiatives.

Comment: A few commenters recommended CMS consider the impact the updated model will

have on one or multiple of the following areas: Health Provider Shortage Areas (HPSA),

Medically Underserved Areas (MUA), high Area Deprivation Index (ADI) locations, high Social

Vulnerability Index (SVI) locations, and/or Opportunity and Enterprise Zones.

Response: CMS thanks the commenters for their recommendations and feedback.

The CMS-HCC model is a national model, including large subgroup segments that capture

national variation in costs between the segmented populations. The goal of risk adjusted

payments is to pay accurately using the appropriate relative risk for a beneficiary. There will be

variation in the impact on risk scores depending on each beneficiary’s clinical mix. As

previously discussed, the model is not intended for provider reimbursement at the service level.

Therefore, the risk adjustment model should not have implications on provider availability.

Regarding vulnerable populations, as previously stated the model is segmented by dual-eligible

status and includes demographic variable to estimate relative risk. Risk scores for dual-eligible

beneficiaries continue to be higher than non-duals on average.

Comment: A commenter wanted to know if the proposed model will use CY 2023 diagnosis

codes.

Response: CMS will use the updated model to calculate risk scores for 2024 payment, using

diagnoses codes from 2023 dates of service.

Comment: One commenter asked CMS to provide logic on the inclusion of some “P” and “Q”

ICD-10 newborn codes within the proposed model.

Response: Certain P and Q codes were moved from nonpayment in the Newborn disease group

to payment in condition-specific disease groups in the updated model. ICD-10 coding guidance

specifies that they may be used regardless of age. P codes are for conditions that originate in the

perinatal period, which may continue throughout a patient’s life. Q codes are for congenital

malformations, deformations, and chromosomal abnormalities, which although present at birth

may not be identified until later in life. In the updated model, these codes are remapped for

clinical cohesion to the respective disease-related HCC in the body system, parallel to codes for

the condition from other ICD-10 chapters (i.e., non-congenital forms, not originating in the

perinatal period). Thus, these P and Q codes are added to the model if the condition warrants

inclusion as payment.

Comment: A few commenters stated that CMS made a technical error in Attachment VI, Table

VI-1 related to the Note on Neurology HCCs and disease interactions.

Response: These commenters are correct. The CY 2024 Advance Notice has a technical error in

the table. Under footnote 2 of Attachment VI, Table VI-1, Neurological is defined as HCCs 108-

192; HCCs 195-199, but it should state HCCs 180-192, 195, 196, 198, 199. This has been

corrected in this document. We appreciate the commenters identifying this.

Comment: Several commenters stated either that the proposed risk adjustment model does not

meet the actuarial equivalence requirement in Section 1853(a)(1)(c)(i) of the Social Security Act

or were concerned that the proposed model did not meet the actuarial equivalence requirement

based on the documentation provided in the Advance Notice. A commenter noted their belief

that in order to fulfill its statutory mandate to ensure actuarial equivalence, CMS must only

assess FFS treatment costs and associated diagnoses when updating the model used for MA

payment, and that the use of other criteria, are not valid unless those changes improve the

model’s ability to predict FFS treatment cost. Commenters identified specific diagnoses that they

stated were excluded from the proposed model, which they believe offers evidence that the

proposed model violates this statutory requirement.

Response: Section 1853(a)(1)(c)(i) of the Social Security Act instructs CMS to adjust the

payments made to MA organizations “for such risk factors as age, disability status, gender,

institutional status, and such other factors as the Secretary determines to be appropriate,

including adjustment for health status . . . so as to ensure actuarial equivalence.” It also

authorizes CMS to “add to, modify, or substitute for such adjustment factors [i.e., age, disability

status, gender, etc.] if such changes will improve the determination of actuarial equivalence.”

The 2024 CMS-HCC model satisfies the statutory requirement of actuarial equivalence because,

within the limitations of such a model, it accurately predicts variations in cost, as discussed at

length above. But the statute does not require CMS to show that every single change made from

the previous CMS-HCC model improves its predictive accuracy. CMS retains the discretion and

authority to design the 2024 CMS-HCC payment model so as to achieve actuarial equivalence,

just as earlier payment models were designed to do. Those earlier payment models do not limit

the scope of the Secretary’s authority today.

The Balanced Budget Act of 1997, the initial statutory requirement to implement risk adjustment

based on health status in the MA program, directed the Secretary to implement a risk adjustment

methodology that accounts for variations in per capita costs based on health status and other

demographic factors no later than January 1, 2000. As required by 1853(a)(3)(A), the risk

adjustment methodology was to be submitted to Congress in a report that included an evaluation

of the method by an outside, independent actuary of the actuarial soundness of the proposal. The

report that was submitted to Congress in 1999 provided a foundation for the ten principles that

were expanded on in a 2000 report to Congress that CMS still uses today to guide risk

adjustment model development.

Each iteration of the risk adjustment model, including the

2024 CMS-HCC model, has maintained or improved upon the previous payment model’s ability

to predict variation in per capita spending across key subgroups of beneficiaries.

CMS is obligated to release a report to Congress every three years with information on how

revisions to the risk adjustment model impact the predictive ratios for groups of enrollees in MA

plans by predicted cost and groups defined by the number of chronic conditions of enrollees.

Because CMS has not uniformly collected cost information for beneficiaries enrolled in the MA

program, the report is developed with data on Medicare FFS enrollees, which are the same data

used to calibrate the model. The next report will be released with findings from the 2024 CMS-

HCC model in 2024. In response to comments from stakeholders, predictive ratios are provided

for the 2024 CMS-HCC model overall on page 72 and by model segment in Attachment VIII

Tables VIII-5 to VIII-11.

One commenter wrongly asserts that “under the policies proposed in the Advance Notice, CMS

would fail to satisfy its obligation to ensure actuarial equivalence, as the proposed model does

not appear to be based on an assessment of FFS costs in order to generate appropriate

coefficients to support appropriate MA payments.” The 2024 CMS-HCC model does account for

FFS cost to generate appropriate coefficients. Per-capita payments to MA plans reflect the

relative cost experience in Medicare FFS for a beneficiary population modeled by reference to

demographics, diagnoses, and other factors CMS selects. As discussed above, the 2024 CMS-

HCC model accurately predicts variation in FFS costs.

One commenter implies that CMS must make decisions to update the risk adjustment model

based only on “analysis of whether removing these codes will mean the model more accurately

predicts costs associated with treating certain conditions in FFS” and erroneously claims that “In

order to fulfill its statutory mandate to ensure actuarial equivalence, CMS must assess FFS

treatment costs associated with diagnoses, not use of codes.” The Secretary is provided

discretion in how to adjust for health status. In addition, as discussed before, the risk adjustment

model accurately predicts variation in per capita cost based on health status as intended.

Diagnosis codes from claims or encounter data are used as a measure of health status to predict

variation in cost because those data are the most practicable for developing and implementing a

payment system at the scale required by the Medicare program. An individual diagnosis code or

See 1999 Report to Congress: https://www.cms.gov/medicare/health-

plans/medicareadvtgspecratestats/downloads/rtc_riskadjusters1999.pdf, and 2000 Report to Congress: Diagnostic Cost Group

Hierarchical Condition Category Models for Medicare Risk Adjustment

For example, see the December 2021 Report to Congress

100

HCC is not a sufficient, nor the only, indicator of a beneficiary’s health status. CMS does not

believe that the reference to health status in Section 1853(a)(1)(C)(i) requires CMS to adjust

payments to MA organizations for the costs they incur for specific diagnoses on submitted

claims or encounter data, in relation to FFS or otherwise. The predictive accuracy for a specific

condition in the model is not a valid criterion on which to assess the model’s actuarial

equivalence because expected relative costs based on health status is ultimately measured by all

conditions and factors in the model used to calculate risk scores.

The statute provides the Secretary discretion in how to account for health status when calculating

payments that reflect the risk of the enrollee, and this extends to the individual HCCs included in

the model. CMS follows the ten principles laid out in the 2000 Report to Congress to group

diagnosis codes into HCCs and select which HCCs are included in the model.

These principles

balance clinical considerations, such as creating clinically meaningful groupings of diagnoses

and ensuring only well-specified diagnoses are included in the model, and the predictive

accuracy of the model as measured by the ability of the model to predict accurately across groups

of beneficiaries. As discussed throughout the Rate Announcement, the 2024 CMS-HCC model

accurately predicts variation in expenditures (per capita cost) across groups of beneficiaries.

There is no meaningful difference in the accuracy of each model for groups of beneficiaries by

decile of predicted risk using the beneficiary sample specifically applicable to each model, CMS’

standard measure of accuracy. However, when compared using the same beneficiary sample (i.e.,

2019 beneficiaries, 2018 diagnoses) for both models, the 2024 CMS-HCC is more accurate by

decile of predicted risk than the 2020 CMS-HCC model. The 2024 CMS-HCC model more

accurately reflects MA plan’s expected experience because it was developed with more recent

data. In general, the model predicts higher expenditures for people who have higher actual

predicted risk and less for people who have lower actual predicted risk. The model accurately

predicts total expenditures regardless of whether or not a condition is included in the model.

Spending associated with conditions not included in the model is allocated to other variables in

the model, such as the demographic factors, count variables and conditions in the model

associated with the removed conditions. To the extent that there are changes in the predicted risk

for certain subgroups of beneficiaries, or beneficiaries with certain conditions, those changes can

occur for multiple reasons. The model includes updates to the underlying data years, the

denominator year, and reclassification for ICD-10. As such, relative changes are driven by more

recent enrollment, treatment/utilization, and cost patterns, in addition to the reclassification.

If CMS were not paying MA organizations appropriately for the health status of the beneficiaries

they enroll, many MA plans would not be competitive with the FFS program and enrollment

would likely stagnate. The exact opposite has occurred. Analysis of CMS program statistics

shows that between 2014 (the first year of the last major clinical revision to the risk adjustment

model) and 2021, enrollment in MA grew by 70 percent whereas enrollment in FFS has declined

2000 Report to Congress: Diagnostic Cost Group Hierarchical Condition Category Models for Medicare Risk Adjustment.

101

by seven percent. Enrollment growth among beneficiaries who are dually enrolled in Medicare

and Medicaid, who on average have higher medical cost and greater health needs, has also been

higher, increasing by more than 125 percent from 2.7 million in 2014 to 6.1 million in 2021,

whereas dual enrollment in FFS has declined by 22 percent from 7.2 million to 5.6 million. Dual

eligible enrollees made up more than one in five (22.1 percent) of MA enrollees in 2021, similar

to the proportion in FFS (19 percent). As enrollment in the program has grown, the number of

organizations offering MA plans has increased, the number of plans offering supplemental

rebates has increased, and plans are profitable. Over the last ten years (2014 – 2023) the average

margin included in plan bids is 4.4 percent while the ratio of the average benchmark to the

average plan bid to provide standard Medicare benefits has increased. In 2014, the adjusted

benchmark was on average 114 percent of the standardized bid, and in 2023 the adjusted

benchmark was on average 152 percent of the standardized bid. The 2024 CMS-HCC model

maintains the same risk adjustment system based on variation in health status that has supported

a robust MA program. The commenters provided no specific evidence in their comments that the

payments they will receive under the 2024 CMS-HCC model are not sufficient to meet their

obligation to provide benefits to the beneficiaries who enroll in their plan.

Comment: One commenter suggested that the documentation standards in Medicare Advantage

differ from the documentation standards in traditional Medicare (i.e., Parts A and B), and that

this difference causes the CMS-HCC risk adjustment model to violate the actuarial equivalence

requirement in Section 1853(a)(1)(c)(i) of the Social Security Act.

Response: The argument, that differing documentation standards between MA and FFS cause the

risk adjustment model to systematically underpay MA organizations in violation of the actuarial

equivalence provision of Section 1853(a)(1)(c)(i) of the Social Security Act, incorrectly

presumes that there are two different standards. As we recently explained, FFS and Medicare

Advantage have the same documentation standard for diagnosis coding. Policy and Technical

Changes to the Medicare Advantage Program for Years 2020 and 2021, Final Rule, 88 Fed. Reg.

6643, 6652-53 n.31, 6657-59 (February 1, 2023). Both programs require all reported diagnoses

to be substantiated by documentation in the beneficiary’s medical records. Moreover, the

commenter is wrong to suggest that MA data is “audited” while FFS data is “unaudited,” or that

coding errors are condoned in FFS. In each program there is i) an identical documentation

standard, ii) an obligation to work in good faith towards compliance with that standard, iii) some

CMS auditing of compliance, and iv) coding errors that nonetheless escape detection. See id. at

6658 n.40, 6659. This has been true since the beginning of the Medicare Advantage program,

and remains true today. There is no reason to presume that the error rate in diagnosis coding in

FFS is meaningfully higher than the error rate in Medicare Advantage.

CMS regularly informs Congress about the use of the CMS-HCC risk adjustment model,

calibrated on FFS data, for calculation of MA payments. Congress has been well informed about

CMS’ implementation of risk adjustment since the advent of the Medicare Advantage program,

often requiring reporting or amending the relevant statutory provisions, without any indication

102

that Congress believed CMS was falling short of the statutory requirement of actuarial

equivalence in payments to MA insurers. To the contrary, Congress has made a holistic

determination that payments to MA organizations based on the CMS-HCC payment model are

higher than they ought to be by at least 5.9%. The statutory adjustment leaves the agency with

asymmetric discretion: CMS is free to find that a larger reduction in payments should be made,

but not a smaller one.

That is not to say that the risk adjustment model achieves perfect accuracy. As noted previously,

the model predicts variation in expenditures for groups of enrollees based on their expected

costs, not expenditures at the individual level.

Nevertheless, accuracy is achieved at the

aggregate level, and as we recently explained, the arguments that FFS errors create systematic

underpayment do not adequately account for the offsetting effects of FFS under-coding and the

increased costs associated with FFS over-coding. 88 Fed. Reg. at 6659.

In fact, most independent experts have found that if anything, MA organizations are significantly

overpaid even after application of the minimum 5.9% coding pattern adjustment that has been in

place since 2019. MedPAC, for example, estimated that MA risk scores for 2020 were 9.5%

higher than FFS risk scores for similar FFS enrollees (3.6% higher after applying the 5.9%

coding pattern adjustment), resulting in nearly $12 billion in excess payments to MA plans.

MedPAC, Report to the Congress: Medicare Payment Policy 440 (March 2022). Other

independent studies have reached similar findings. See id. at 440-41 (citing studies). Further,

some independent experts estimate that these differentials are increasing. Very recently,

MedPAC estimated that for 2021, MA risk scores were 10.8% higher than FFS risk scores for

similar FFS enrollees (4.9% higher after applying the 5.9% coding pattern adjustment), resulting

in nearly $17 billion in excess payments to MA plans.

MedPAC projects that excess payments

to MA plans will reach $23 billion in 2023 if MA coding patterns remains the same as in 2021.

CMS is required to operate the MA program and publish payment rates each year. The agency

must therefore act on the basis of the best information available to it at the time, within the

bounds of the discretion afforded to it by Congress; it cannot wait for more complete

information. The history of the CMS-HCC payment model, which has relied on the same

documentation standard and been calibrated on the same data for many years; the amendments to

the Medicare statute indicating a congressional determination that the model was paying MA

See UnitedHealthcare, 16 F.4th at 874 (“The model uses data from a large pool of beneficiaries (full sample sizes over 1 million for the CMS-

HCC models) to estimate predicted costs on average for each of the component factors (e.g., age-sex, low income status, individual disease

groups). Using regression analysis on such a vast data sample mutes the effect of individual errors in traditional Medicare data, so long as errors

are not so widespread or systemically skewed as to raise or lower the values of particular relative factors.”) (internal citations and quotation

marks omitted).

See also UnitedHealthcare, 16 F.4th at 888 (discussing the effects of underreporting and extraneous expenditures within the

FFS data).

MEDPAC, REPORT TO THE CONGRESS: MEDICARE PAYMENT POLICY xxiv, 324-25, 355 (March 2023); LUIS SERNA & ANDY

JOHNSON, MEDPAC PRESENTATION Slides 11, 13 (Jan. 12, 2023), https://www.medpac.gov/wp-

content/uploads/2023/01/MedPAC-MA-status-report-Jan-2023.pdf.

Id.

103

organizations more than they ought to be paid; and the weight of expert analysis suggesting that

the model continues to overcompensate those insurers despite the mandatory payment reduction,

all demonstrate that CMS acts within its statutory authority in continuing to operate the risk

adjustment model on FFS cost and diagnosis data while making only the statutory minimum

payment reduction.

Comment: A commenter challenged CMS authority to implement changes to Medicare

Advantage payment policy through the Advance Notice and Rate Announcement. Several

commenters contested CMS authority to make any changes to the Part C risk adjustment model

generally. The commenters argued that what CMS proposes makes fundamental changes to the

MA program that are unprecedented and therefore exceeding the executive branch’s authority to

implement, and that the proposed changes should be implemented through notice and comment

rulemaking.

Response: The commenter mischaracterizes the proposed changes to payment in the 2024

Advance Notice. The proposed changes are not unprecedented. They are ordinary and routine.

The fundamental structure of the Medicare Advantage program remains intact. CMS will

continue to pay MA plans a risk adjusted amount based on the MA plan’s bid to provide

Medicare Parts A and B benefits through adequate provider networks; in addition, for MA plans

that bid under the benchmark, CMS will provide the MA plan a portion of the amount by which

the benchmark exceeds the bid as the MA rebate, to be used to furnish supplemental benefits

(including reductions in cost sharing) or to pay Part B or Part D premiums for their enrollees.

CMS pays MA organizations rebates to cover additional benefits as an incentive for the MA

organization to provide Part A and Part B services for less than the cost to the FFS program, as

measured by the MA benchmark applied in each plan’s bid. By statute, the plan-specific rebate

amount is based on the MA contract’s star rating and the MA organizations ability to cover Parts

A and B benefits for less than the applicable benchmark rate. This Advance Notice and Rate

Announcement does not alter the payment structure; rather it updates the benchmark rates and

other aspects of MA payment calculation.

CMS expects the net impact to aggregate MA payments between 2023 and 2024 as a result of all

changes finalized in this Rate Announcement to increase MA payments by 3.32 percent.

CMS

anticipates stable premiums and benefits for beneficiaries in 2024, as seen previously in years

with comparable updates.

The updates made to the CMS-HCC Part C risk adjustment model as finalized in the 2024 Rate

Announcement are not unprecedented. They are similar in scope to risk adjustment model

HHS and the GAO have also both identified high amounts of improper payments in the MA program. 88 Fed. Reg. at 6645,

6653. In fiscal year (FY) 2021 for example (based on calendar year 2019 payments), we calculated that CMS made over $15

billion in Part C overpayments, a figure representing nearly 7 percent of total Part C payments. Id. at 6645 n.7 (citing HHS, FY

2021 HHS Agency Financial Report, pg. 211. See https://www.hhs.gov/sites/default/files/fy-2021-hhs-agency-financial-

report.pdf

104

changes proposed and finalized previously through the Advance Notice and Rate Announcement.

The 2024 CMS-HCC model changes are a combination of routine data updates (e.g., updating

the years of data used when calibrating the model) and clinical updates to the Hierarchical

Condition Categories (HCCs) that were required to develop a model using the ICD-10 diagnosis

codes implemented in 2015. For example, CMS recalibrated the MA risk adjustment model with

updated data years and/or made clinical revisions to the risk adjustment model in payment years

2007, 2009, 2013, 2014, 2017, 2019, and 2020. The 2021 Report to Congress provides the

history of the models and changes over the years. A major clinical revision, including the

removal of HCCs for dementia based on Principle 10, was completed for the models that were

implemented for the Program of All-Inclusive Care for the Elderly (PACE) starting in PY 2012

and phased in for MA starting in PY 2014. Updates to the data years–pairwise years of diagnosis

and cost information–to adjust the model coefficients to account for more recent patterns of

health status and cost in the FFS Medicare program are more frequent and last occurred with

model update for PY 2019.

Although requested and suggested by commenters, CMS is not required to use rulemaking in the

Federal Register to adopt a new risk model for MA payment. Section 1853(b)(2) directs that

CMS determine and announce the annual MA capitation rate and the risk and other factors to be

used in adjusting such rates under section 1853(a)(1)(C) for MA payments in each contract year.

Each year, in advance of the final announcement of the rates and risk and other factors, CMS

must “provide for notice to [MA] organizations of proposed changes to be made in the

methodology from the methodology and assumptions used in the previous announcement and

shall provide such organizations an opportunity (in 2017 and each subsequent year, of no less

than 30 days) to comment on such proposed changes.” CMS fully complied with that required

process: the CY 2024 Advance Notice described the changes in methodology and assumptions

for setting the capitation rates and described how CMS developed the risk factors and specific

coefficients for adjusting those capitation rates for age, disability status, gender, institutional

status, eligibility for Medicaid, and health status. This notice and comment process has been in

place since 1985 when CMS first began contracting with private health plans on a capitation

basis, under procedures set forth in section 1876(a)(1)(F) of the Act. All major changes in

payment policy have been implemented through this process. For example, when section

1853(a)(3) was first implemented in 2000 with the initial risk adjustment methodology

developed by CMS, this initial methodology was implemented through an Advance Notice and

Rate Announcement. All subsequent changes to the risk adjustment methodology, including the

establishment of a budget neutrality factor to make risk adjustment payments budget neutral to

the prior demographic-based payments, used in the initial years of CMS-HCC model, and the

subsequent decision by CMS to phase out budget neutrality (which was ratified by Congress in

the DRA) have all been implemented through the section 1853(b) process. Congress has on

several occasions ratified in statute methodologies that CMS established through this section

1853(b) process (e.g., the initial phase in of risk adjustment, the plan to phase out budget

neutrality, providing separate adjustments in the risk adjustment model for dual eligible

105

individuals). Furthermore, section 1853(b)(2) requires only a 30-day comment period, which we

believe is an indication that this period is sufficient for stakeholders to review the changes

proposed in the Advance Notice and provide comment. Prior to the Securing Fairness in

Regulatory Timing Act of 2015 (SFRTA) (Pub. L. 114-106), which amended section 1852(b) to

require that the Advance Notice be issued 60 days before the issuance of the Rate

Announcement, the statute did not require that the comment period be a minimum length.

Although the 21st Century Cures Act required, at section1853(a)(1)(I)(iii), a 60-day comment

period for changes to the risk model under section 1853(a)(1)(I), that comment period does not

apply to future model updates made by the Secretary under the authority granted in Section

1853(a)(1)(C).

Comment: Several commenters stated that they are directionally aligned with CMS’ decision to

update the risk adjustment model based on more current underlying data, updating the

denominator, and some clinical revisions based on ICD-10, but were specifically opposed to the

clinical revisions to select conditions made under Principle 10.

Response: Commenters conflated changes to the model that were exclusively based on Principle

10 and changes that were based on Principle 10 in conjunction with other model principles. The

transition to ICD-10 required CMS to reevaluate all conditions (those ICD-10 diagnoses that

were included in the mapping to payment HCCs in the current 2020 CMS-HCC model as well as

all diagnoses not mapped to payment HCCs in the 2020 CMS-HCC model) against all of the

model principles. Between the 2020 CMS-HCC model and the 2024 CMS-HCC model, 2,236

ICD-10 diagnosis codes no longer map to the model for payment and 209 were added.

Additional information for the 2,236 codes that newly map to non-payment HCCs in the 2024

CMS-HCC model is provided below. These diagnosis codes generally fall into one of six

categories:

1. Subsequent Encounter (codes ending in D) – 6 percent of the diagnosis codes that newly

map to non-payment HCCs under the 2024 CMS-HCC model. Subsequent encounter

codes represent less intensive follow-up treatment. While the cost from subsequent

encounters are included in the model, the diagnosis codes indicating they are from a

subsequent encounter were removed because they do not reliably or consistently predict

meaningful costs. The diagnosis code on the initial encounter is more reliable and

predictive. Costs from these encounters will be attributed to other conditions on the

encounter that are in the model, demographic factors and count variables.

2. Sequela (codes ending in S) – 40 percent of the diagnosis codes that newly map to non-

payment HCCs under the 2024 CMS-HCC model. The initial encounter diagnosis codes

for these conditions that are being actively treated and predict ongoing and prospective

costs still map to payment HCCs. Sequela codes are for late effects of a condition and

indicate variable conditions that can be acute and/or require less follow up medical care.

Because ICD-10 guidelines require the actual late effect (e.g., pain) to be coded

separately, the sequela codes, which are non-specific, are excluded from payment HCCs.

106

Costs from these encounters will be attributed to other conditions on the encounter that

are in the model, demographic factors and count variables.

3. Drug-induced – 8 percent of the diagnosis codes that newly map to non-payment HCCs

under the 2024 CMS-HCC model. ICD-10 codes for drug-induced conditions can be

secondary to poor quality of care related to drug-induced conditions or drug-intensive

styles of care. Re-mapping these conditions to a non-payment HCC focuses the costs on

underlying health status risks rather than the side effects of health care, and because drug-

induced conditions are often temporary/reversible and are unlikely to predict prospective

costs (i.e., the condition is often resolved by stopping the drug). Beneficiaries with a

persisting drug-induced condition are ultimately coded with a diagnosis describing their

condition that is not labeled as “drug-induced.” Thus, expenditures associated with drug-

induced conditions will be reflected in costs modeled for the underlying health condition

prompting drug treatment. For example, costs of chemotherapy and its complications will

load onto associated cancer diagnoses. Only a few codes (16 unique codes and the drug

or chemical induced diabetes E09- code set) were removed from payment primarily

because they are drug-induced. Other drug-induced codes were made nonpayment both

because they are drug-induced and due to another clinical/empirical consideration (for

example, overprediction of the costs for beneficiaries with those conditions in the 2020

CMS-HCC model, clinical considerations based on consulting physicians’

recommendation, or the non-drug-induced version of the condition was also moved to

nonpayment).

4. Complication of Medical Care – 16 percent of the diagnosis codes that newly map to

non-payment HCCs under the 2024 CMS-HCC model. The goal of risk adjustment is to

use health status to predict the relative additional costs of treating the beneficiary, and not

the side effects of medical treatment, which are not reliable predictors of future costs. For

example, in the 2024 CMS-HCC model, relative costs associated with an infection as a

complication of care will load onto the underlying condition prompting the medical care.

Claims data do not allow for determining whether a complication is avoidable. Excluding

complications, such as a surgery performed on the wrong side of the body, avoids using

diagnoses that result from poor quality of care, or more aggressive, procedure-oriented

care. Serious complications would be accounted for by the specified conditions that are

included in the payment model, which would be coded in addition to or instead of

complications codes, especially if they persist in the long-term.

5. Principle 10 only – 3 percent of the diagnosis codes that newly map to non-payment

HCCs under the 2024 CMS-HCC model. These diagnosis codes were removed after

clinical assessment, conducted after the codes were identified for additional clinical

review. The list of HCCs identified for further review is in the subsequent comment

summary and response. As noted before, the clinical panel provided insight and

recommendations related to the criteria used to evaluate and treat conditions (e.g., lab

findings, test results, physical exams, medications, medical interventions, etc.);

107

specificity of diagnoses; potential for variable coding for certain diseases or disease

groups; treatment practices; manifestations of conditions in different populations (e.g.,

aged vs. disabled; non-dual vs. full-benefit; dual vs. partial-benefit dual); disease severity

and implications for medical burden; clinical face validity (i.e., whether the condition is

consistent with the relative costs associated with it) and interpretation of empirical

results; clinical similarities and differences between specific diseases (e.g., Crohn’s

disease and ulcerative colitis); relationship of acute and chronic versions of diseases (e.g.,

acute hepatitis C and chronic hepatitis C); and implications for predicted costs.

6. Other – 28 percent of the diagnosis codes that newly map to non-payment HCCs under

the 2024 CMS-HCC model. This category indicates decisions were made based on a

combination of clinical and empirical reasons that align with model principles such as

clinical meaningfulness (Principle 1), overprediction (Principle 2), diagnostic specificity

(Principle 5), and concerns about coding intensity or clinical discretion (Principle 10.)

As can be determined from the distribution of the reasons for exclusion from the 2024 model, a

significant majority of diagnosis codes (97 percent) were excluded for reasons other than

Principle 10. As noted in the 2024 Advance Notice Fact Sheet, analysis of the most recent data

suggests fully implementing the 2024 CMS-HCC model would reduce MA risk scores by 3.12

percent. A majority of the -3.12 percent impact is due to changes other than those based only on

Principle 10. We calculate the marginal impact of the changes based only on Principle 10 to

be -0.55 percent (e.g., Principle 10 only changes account for -0.55 of the total -3.12 percent

impact).

A list of conditions that commenters were concerned about excluding diagnosis codes from and

the rationale for removing those diagnosis codes is provided in response to other comments in

this section.

Comment: Several commenters stated that the changes to the Part C risk adjustment model

proposed in the 2024 Advance Notice based solely on Principle 10 are intended to reduce

payments to MA organizations because of differences in coding patterns between Medicare

Advantage and FFS and therefore are duplicative of the coding pattern adjustment. Commenters

indicated those changes are not authorized by section 1853(a)(1)(C) of the Social Security Act

and one commenter noted that such changes would require CMS to implement a model based on

MA encounter data.

Response: We disagree. Section 1853(a)(1)(C)(ii) of the Social Security Act requires the

Secretary to reduce payments to MA organizations by at least 5.9% to account for “differences in

coding patterns between Medicare Advantage plans and providers under part[s] A and B.” As

explained above, per-capita payments to MA plans reflect the relative cost experience in

Medicare FFS for a beneficiary population modeled by reference to demographics, diagnoses,

and other factors CMS selects. The per-capita payments initially calculated by the CMS-HCC

model are then reduced in compliance with section 1853(a)(1)(C)(ii). But the enactment of that

108

provision did not diminish the Secretary’s discretion or authority to design the payment model.

And as discussed above, the 2024 CMS-HCC model accurately predicts variation in FFS costs,

just as the 2020 CMS-HCC model did when it was published.

The removal of certain codes from payment HCCs and constraints on the coefficients of certain

HCCs are not the same as the coding pattern adjustment required by section 1853(a)(1)(C) and

do not address the same thing. The adjustment mandated by section 1853(a)(1)(C)(ii) is applied

to each beneficiary’s risk score at every MA plan, which reflects that provision’s focus on coding

pattern differences at a general, program-wide level. Specifically, section (a)(1)(C)(ii) addresses

the incentive for all MA plans to identify more valid, supported codes on all HCCs, and therefore

report more diagnoses (which we do not assume to be inaccurate when evaluating coding pattern

differences between FFS and MA). By reducing each individual risk score by a specified

percentage, the coding pattern adjustment corrects for overall, net differences between MA and

FFS coding. Our process in applying Principle 10 and looking at how often certain codes

appeared within the universe of diagnosed conditions for Medicare beneficiaries in FFS and in

the MA program did not look at general, program-wide differences on all HCCs, but rather at

whether certain differences on particular codes warranted further review to ensure that they are

credible predictors of variation in per capita costs. In circumstances where there is discretion in

coding, beneficiaries with different profiles are attributed with the same HCC and the model’s

ability to differentiate higher-than-average relative costs accurately is diminished.

Principle 10 is well grounded and appropriate to use because it focuses on ensuring that codes

used in the model are clinically and empirically credible predictors of future costs. Because the

coding of additional diagnoses directly increases the payments made to MA organizations,

diagnoses for which the documenting and reporting involves a substantial amount of discretion

present a particular risk to the integrity of the payment model.

This risk is distinct from the

more general incentive, which is addressed by Section 1853(a)(1)(C)(ii) of the Social Security

Act, for MA organizations to increase their coding intensity (and therefore their payments) by

reporting as many valid, supported diagnoses as possible.

The model relies on conditions that are well specified, and that consistently and reliably predict

medical costs in a future year. Incentives for discretionary coding, while possibly stronger in

MA, are not unique to MA, and we have seen increases in coding in the FFS program in recent

years. There are several possible drivers. The proportion of payments in FFS tied to value-based

payment models that rely on the CMS-HCC model has grown significantly between the

expenditure year for the 2020 model (2015) and the expenditure year for the 2024 CMS-HCC

model (2019). Further, providers who are both FFS participating providers and in MA networks

may carry over incentives from the MA program into their larger practice. Including diagnosis

MedPAC, Report to the Congress: Medicare Payment Policy xxiv, 324-25, 355 (March 2023); Luis Serna & Andy Johnson,

MedPAC Presentation Slides 11, 13 (Jan. 12, 2023),

https://www.medpac.gov/wp-content/uploads/2023/01/MedPAC-MA-status-report-Jan-2023.pdf

109

codes in the model where providers may have an incentive to report a more severe diagnosis

code when clinically not well specified dilutes the cost predicted by the model for more severe

conditions and potentially reduces the sample size of the less severe condition resulting in

unstable predictions between calibrations. As described before, despite including fewer diagnosis

codes, the reclassified payment model maintains or improves predictive accuracy compared to

the model currently in use (the 2020 CMS-HCC model) predictive ratios (ratios of predicted to

actual mean expenditures by predicted expenditure decile) and other measures CMS does not

typically consider, such as the R

statistic.

Further, even if there was an overlap between the Principle 10-guided changes in the 2024 CMS-

HCC model compared to the currently used 2020 CMS-HCC model, nothing in section

1853(a)(1)(C) prohibits how CMS has approached this issue in model development. Section

1853(a)(1)(C)(i) authorizes the addition, modification, or removal of a risk factor. The

mandatory, uniform coding pattern adjustment did not (as some commenters suggested) remove

the Secretary’s authority to revise the model to eliminate or constrain particular codes or HCCs

based on our longstanding principles for risk model development for the MA program. Here, the

revisions to address the multiple conditions (as reflected in coded diagnoses) and costs used by

CMS to adjust for overall health status are within the scope of that authority. We note that

commenters alleging duplication with the coding pattern adjustment provided no specific

evidence in their comments that they will be underpaid. There is a range of estimates for the

appropriate coding pattern adjustment, and as noted before, independent experts have found that

if anything, the risk adjustment model has been overpaying MA organizations significantly even

after application of the minimum coding pattern adjustment.

The changes finalized in the 2024 CMS-HCC model are incorrectly characterized by the

commenters as a coding intensity or coding pattern adjustment. The changes based on Principle

10 are clinically informed and are intended to ensure a more accurate prediction of variation in

cost based on health status. Removing HCCs for protein-calorie malnutrition, angina pectoris,

and atherosclerosis of arteries of the extremities with intermittent claudication, and separately

constraining the three diabetes HCCs (glycemic, unspecified, or no complications; with chronic

complications; and with severe acute complications) and congestive heart failure (heart failure,

except end-stage and acute; acute heart failure (excludes acute on chronic); and acute on chronic

heart failure) so that each grouping has the same coefficient allocates the FFS spending

associated with those conditions to demographic factors, the count variables, and other

associated conditions in the model that are better specified.

As noted in the Advance Notice, CMS followed a process for identifying ICD-10 codes that may

be subject to discretionary coding. Discretionary coding can occur when providers may make

For a discussion of different methods for calculating a coding intensity adjustment see, “Reducing Medicare Advantage

Overpayments” at https://www.crfb.org/papers/reducing-medicare-advantage-overpayments. See also, the 2010 Advance Notice

https://www.cms.gov/medicare/health-plans/medicareadvtgspecratestats/downloads/advance2010.pdf

110

different decisions regarding the diagnosing and documenting of a condition given the same

circumstance, or an MA organization may make a decision that is different from a provider

regarding the reporting of condition for payment. This may occur because clinical indicators are

broad, need significant interpretation, or because the condition is being diagnosed and

documented in situations where it has no clinical significance, or where it does not require or

affect patient care, treatment or management as required by the ICD-10 Coding Guidelines. The

first step in our analysis was to determine whether there are differences in coding patterns

between FFS and MA by condition, because a difference in the rate of diagnosis codes submitted

could be indicative of differences in clinical interpretation. CMS data and independent studies

generally find that the demographic and healthcare utilization patterns between FFS and MA are

similar.

27,28,29

By and large, the prevalence of HCCs is also similar between MA and FFS. In the

Non-Dual Aged segment, the percentage point difference in prevalence between MA and FFS

was within 1 percentage point for 75 of 86 payment HCCs (87 percent). In the Full Benefit Dual

Aged Segment, the percentage point difference in prevalence between MA and FFS was within 1

percent for 72 of the 86 (84 percent) payment HCCs. Therefore, the small number of outlier

HCCs where the difference in prevalence in MA exceeded 1 percent warranted further analysis.

CMS identified the following HCCs in the 2020 CMS-HCC model that had significant variation

in coding between MA and FFS:

1. HCC 18 Diabetes with Chronic Complications

2. HCC 21 Protein-Calorie Malnutrition

3. HCC 22 Morbid Obesity

4. HCC 59 Major Depressive, Bipolar, and Paranoid Disorders

5. HCC 85 Congestive Heart Failure

6. HCC 88 Angina Pectoris

7. HCC 108 Vascular Disease

8. HCC 111 Chronic Obstructive Pulmonary Disease

9. HCC 138 Chronic Kidney Disease, Moderate (Stage 3).

Identifying HCCs with significant coding pattern differences was the first step in CMS’ analysis

before a more intensive clinical review to determine whether the continued reliance on these

diagnosis codes in the payment model was consistent with Principle 10.

Among diagnoses in these nine HCCs that were selected for further clinical review, CMS

determined that the following ICD-10 diagnoses codes are subject to sufficient clinical discretion

https://www.commonwealthfund.org/publications/issue-briefs/2021/oct/medicare-advantage-vs-traditional-medicare-

beneficiaries-differ; see also Paul D. Jacobs & Richard Kronick, Getting What We Pay For: How Do Risk-Based Payments to

Medicare Advantage Plans Compare with Alternative Measures of Beneficiary Health Risk, 53 H

EALTH SERVICES RESEARCH

4997, at 5012 (Dec. 2018) (increasing MA risk scores compared to FFS likely outpace relative increases in actual health risk; MA

enrollees are no sicker, and may well be healthier than similar FFS beneficiaries).

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9175080/

https://www.kff.org/medicare/report/beneficiary-experience-affordability-utilization-and-quality-in-medicare-advantage-and-

traditional-medicare-a-review-of-the-literature/

111

and that removal from payment HCCs in the 2024 CMS-HCC model is appropriate: all ICD-10

diagnosis codes mapped to 2020 CMS-HCC model HCC 21 Protein-Calorie Malnutrition and

HCC 88 Angina Pectoris, and 30 diagnosis codes mapped to HCC 108 Vascular Disease. CMS

further assessed that the coefficients in the 2024 CMS-HCC model for the diabetes HCCs and

some of the Congestive Heart Failure HCCs, which retain most of the ICD-10 codes originally

mapped to the 2020 CMS-HCC model HCCs, should be constrained. Constraining the

coefficients for the diabetes HCCs in the 2024 CMS-HCC model balances the need to predict

cost for clinically meaningful conditions and the need to limit the incentive to report

discretionary codes included in the model for payment. As a group, each set of constrained

HCCs are accurately predicted, although individual HCCs within each group may be under or

over predicted.

The decision to exclude diagnoses was made after careful consideration and consultation with

independent clinicians. Diagnoses mapped to HCC 21 Protein-Calorie Malnutrition had been

included in the 2020 CMS-HCC model because clinically, protein-calorie malnutrition can be

useful as a risk marker for frailty, severe illness, and late- or end-stage disease, and it is likely

correlated with poverty/disadvantaged populations. However, CMS analysis confirmed that the

most commonly reported diagnosis codes mapped to HCC 21 by FFS and MA include less

severe or unspecified protein-calorie malnutrition (E46, E43, E440) and Cachexia (R64). Recent

analysis from the HHS Office of the Inspector General (OIG) found that hospitals correctly

billed FFS Medicare for severe malnutrition diagnosis codes (E41 and E43) for 27 of the 200

claims reviewed. For 164 of the remaining claims OIG found that hospitals used severe

malnutrition diagnosis codes when they should have used codes for other forms of malnutrition

or no malnutrition diagnosis code at all, and for nine of these claims, the medical record

documentation supported a secondary diagnosis code other than a severe malnutrition diagnosis

code.

This finding in conjunction with clinician experts’ opinion suggested that including even

the more severe diagnoses mapped to HCC 21 in the 2020 model would create incentives for

wide variation in coding. Because the regression model predicts all FFS costs, any spending

associated with the Protein-Calorie Malnutrition HCCs will be reallocated to correlated

conditions or demographic markers, such as cancer and older ages.

As previously stated, the

risk adjustment model is intended to differentiate relative risk such that higher payments are

calculated for sicker beneficiaries with higher disease burden relative to average.

Diagnoses mapped to HCC 88 Angina Pectoris in the 2020 CMS-HCC model were excluded due

to Principle 10 because of a lack of clinical specificity. Decisions regarding whether to diagnose

and document angina pectoris are subject to some discretion, which presents a risk that the

https://oig.hhs.gov/oas/reports/region3/31700010.pdf.

CMS has explained this type of effect before: “We do note that when specified HCCs are removed from the model, the model

is recalibrated and the same costs are predicted with the new set of HCCs. The relative factors for conditions that are comorbid

with the excluded HCC may increase, as may the various demographic factors.” CY 2014 Rate Announcement at 30-31 (April 1,

2013), https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Downloads/Announcement2014.pdf

112

diagnosis could be used in cases of chest pain that do not require further treatment, and also

when the diagnosis does not affect the care, treatment or management of the beneficiary.

Additionally, HCC 88 included ICD-10 codes for other or unspecified forms of angina pectoris,

including with atherosclerosis. Independent studies have found that incidental detection of

atherosclerosis through CT scans is “relatively prevalent and carr[ies] high risk for asymptomatic

coronary disease.” For example, in a study of 1,494 clinically healthy adults without a history of

CVD, 68% of those who underwent abdominal CT had atherosclerotic lesions with stenosis in

the aorta or its major branches. Thus, there is a likelihood that atherosclerosis is being coded

based on radiology reports for cases that are asymptomatic.

This rationale was also applied to

the ICD-10 diagnosis codes previously included for payment in HCC 108 in the 2020 CMS-HCC

model. Arterial Atherosclerosis with intermittent claudication may not be a good predictor of

medical expenditure because evidence suggests that patients with intermittent claudication often

do not require treatment.

In response to a comment suggesting that variation in coding between MA and FFS must be

addressed through a model calibrated with MA encounter data, CMS agrees that an MA coding

pattern difference adjustment would no longer need to be applied to the MA risk scores if the

risk adjustment model were calibrated with MA data. While CMS disagrees with commenters

that changes to the 2024 CMS-HCC model based on Principle 10 are duplicative of the coding

intensity adjustment, and disagrees that CMS does not have the authority to make changes to the

risk adjustment model to account for variability in diagnosis codes submitted, CMS agrees that

the statute supports calibrating a model using MA data submitted on encounter data records.

CMS will engage with stakeholders when we move forward with developing such a model.

CMS-HCC Risk Adjustment Model for PACE organizations for CY 2024

Comment: Several commenters opposed CMS’ proposal to use the 2017 CMS-HCC risk

adjustment model for payment to PACE organizations for CY 2024. These commenters urged

that, for CY 2024, CMS transition PACE organizations to the CMS-HCC risk adjustment model

that is currently being used for MA payment (the 2020 CMS-HCC risk adjustment model). These

commenters are concerned that the 2017 CMS-HCC risk adjustment model excludes several

chronic conditions, especially dementia, which is prevalent in the PACE population.

Commenters also pointed to pressure ulcers (HCC 159), moderate chronic kidney disease (HCC

138), and several mental health and substance use disorder HCCs as condition categories that

they believe are critically important for accurate risk adjustment for PACE organizations.

Commenters believe that the frailty adjustment is inadequate to capture the risk of their

Suh, B., Song, Y. S., Shin, D. W., Lim, J., Kim, H., Min, S. H., ... & Cho, B. (2018). Incidentally detected atherosclerosis in

the abdominal aorta or its major branches on computed tomography is highly associated with coronary heart disease in

asymptomatic adults. Journal of cardiovascular computed tomography, 12(4), 305-311.

Cassar K. (2006). Intermittent claudication. BMJ (Clinical research ed.), 333(7576), 1002–1005.

https://doi.org/10.1136/bmj.39001.562813.DE

113

participants in a model that excludes the aforementioned conditions, which are included for

payment in the 2020 CMS-HCC risk adjustment model. Commenters asserted that PACE

organizations are underpaid using the current risk adjustment methodology because they believe

it fails to recognize the costs of care associated with a complex population, such as PACE

participants. One commenter stated that their ability to provide integrated, quality care to PACE

participants may be compromised because of systematic underpayments they believe are present

for PACE organizations due to the use of the 2017 CMS-HCC risk adjustment model.

Additionally, commenters voiced concern over the misalignment of PACE and MA in regards to

the risk adjustment models used for payment, stating that the misalignment is further exacerbated

by CMS’ proposal to move MA plans to an updated CMS-HCC risk adjustment model that is

calibrated using ICD-10 diagnoses, as well as more recent diagnostic and cost data, thereby

further increasing health disparities. The commenters agreed with CMS that it is important to

update the diagnostic and cost data on which the risk adjustment model is based, and noted that

the model proposed for organizations other than PACE has new categories that reflect more

clinical specificity and validity (i.e., greater level of detail that allows more precision in the

identification of specific conditions). Commenters questioned why these considerations are not

being applied to the risk adjustment model being used for PACE, believing that CMS should use

a payment model for PACE that accurately represents the population they serve.

Generally, commenters believed that it is possible for CMS to use the 2020 CMS-HCC risk

adjustment model to pay PACE organizations, without having to change the diagnostic data

sources currently being submitted by PACE organizations for risk adjustment purposes (which

includes diagnoses submitted to the Risk Adjustment Processing System (RAPS) and the

Encounter Data System (EDS)). Commenters believed that CMS’ past justifications for being

unable to move PACE to the 2020 CMS-HCC risk adjustment model were insufficient or not

credible. Commenters noted that in the 2020 Rate Announcement,

CMS stated that they have

done extensive work to improve and facilitate the submission of encounter data and will continue

to do so, however, commenters were only aware of a couple of instances where CMS engaged

with PACE organizations since that time (site visits with some PACE organizations and a PACE-

specific webinar presentation on encounter data). Commenters argued that CMS has used the

data submitted to RAPS and EDS for PACE risk adjustment for several years “without concern

for accuracy or completeness,” and, therefore, CMS’ prior justifications for not moving PACE to

the model used for MA is not a credible barrier to transitioning PACE to a more accurate risk

adjustment model. They recommended that CMS transition PACE to the 2020 CMS-HCC risk

adjustment model using the current method of calculating PACE risk scores (pooling RAPS,

encounter data, and FFS claims), and that during the transition, CMS collaborate with PACE

organizations to address encounter data submission difficulties they may be experiencing.

CY 2020 Rate Announcement.

114

Response: CMS will continue the use of the 2017 CMS-HCC risk adjustment model for risk-

adjusted payment to PACE organizations for CY 2024. The 2017 CMS-HCC risk adjustment

model was first adopted for PACE in the 2022 Rate Announcement.

CMS acknowledges

concerns from commenters about the inclusion of dementia and other conditions in the model,

and the recommendation to move PACE to the 2020 CMS-HCC risk adjustment model that

includes these conditions. CMS recognizes that using distinct HCCs to calibrate separate models

for PACE and MA may result in differences in predicted risk for individual beneficiaries,

however, we note that the costs associated with conditions that are not in the 2017 CMS-HCC

risk adjustment model for payment, such as dementia, are predicted by comorbid conditions and

demographic factors. To the extent that these costs are not predicted by the model, they are

reflected in the frailty factors.

As noted in the November 1, 2013 HPMS memo titled, “Clarification to Encounter Data

Submissions Memo for PACE Organizations,”

PACE organizations are only required to submit

encounter data records for Medicare-covered items and services for which the organization

collects claims. Because PACE organizations are not submitting encounters for all of their

services, we do not have complete encounter data for their members and, therefore, we do not

have a complete diagnostic profile for their members in the encounter data. Without a complete

diagnostic profile, we cannot rely solely on encounter data to calculate PACE risk scores and,

instead, use diagnoses from encounter data as a supplement to RAPS data when calculating risk

scores for payment using the 2017 CMS-HCC risk adjustment model. Because the 2020 CMS-

HCC risk adjustment model was calibrated using FFS diagnoses that were selected using the

filtering method that is used for encounter data, this model is intended to calculate risk scores

using diagnoses submitted on encounter data records and FFS claims (for beneficiaries who

switch from FFS to MA) filtered in the same manner as encounter data records. Since we do not

have complete encounter data from PACE organizations, we are not calculating PACE

beneficiary risk scores using diagnoses solely from encounter data and FFS claims (in contrast to

the long-standing approach to calculating non-PACE beneficiary risk scores), and we cannot

implement the 2020 CMS-HCC risk adjustment model for PACE at this time. Using primarily

RAPS-based diagnostic data (based on specialty filtering) on a model that was developed to

calculate risk scores using encounter data (based on CPT/HCPCS filtering) results in inaccurate

risk scores, and therefore payment.

In the summer of 2022, CMS engaged with some PACE organizations to discuss successes and

challenges they have experienced with submitting encounter data. CMS will conduct analyses to

ascertain the state of encounter data submissions for PACE organizations and continue to work

closely with these organizations to develop further guidance and provide technical assistance

with transitioning PACE organizations fully to encounter data in anticipation of future

implementation of a risk adjustment model for PACE that is calibrated using encounter data

CY 2022 Rate Announcement

Clarification to Encounter Data Submissions Memo for PACE Organizations, November 1, 2013

115

(such as the one currently used for MA payment in 2023 or a future model). As noted in the 2022

Rate Announcement, CMS continues plans to use a more recently updated model to pay PACE

organizations as soon as it is practicable.

Section K. End Stage Renal Disease (ESRD) Risk Adjustment Models for CY 2024

CMS did not receive comments on the CMS-HCC ESRD risk adjustment models for PACE

organizations for CY 2024. CMS will continue to calculate risk scores for payment of

beneficiaries with ESRD in PACE organizations using the CY 2019 CMS-HCC ESRD risk

adjustment models as proposed in the CY 2024 Advance Notice.

Comment: Two commenters supported the continued use of the CY 2023 CMS-HCC ESRD

Models for 2024.

Response: CMS appreciates the comments. For CY 2024, we will continue to calculate risk

scores for payment of beneficiaries with ESRD in MA plans and certain demonstrations using

the CY 2023 CMS-HCC ESRD risk adjustment models as proposed in the CY 2024 Advance

Notice.

Comment: A couple commenters made recommendations for CMS to engage with stakeholders

when updating the ESRD risk adjustment models in the future; one commenter specifically

requested full transparency, collaboration with plans and providers, and a longer comment period

(i.e., at least 60 days for comment on any proposed changes). A commenter urged that risk

adjustment model changes be finalized at least two years in advance of the implementation date

to allow sufficient time for plans and providers to make necessary operational changes.

Response: Thank you for your comment.

Comment: One commenter requested that CMS move enrollees with a reported dialysis

diagnosis into the ESRD segment automatically.

Response: We believe the commenter is suggesting that CMS use ICD-10 diagnosis codes

related to ESRD as an indicator to calculate risk scores using the ESRD dialysis risk adjustment

model. The ESRD dialysis model is intended to calculate projected risk for beneficiaries

receiving ongoing dialysis treatment. ICD-10 ESRD diagnosis codes are not always indicative of

dialysis treatment. For example, ESRD ICD-10 diagnosis codes can be used in instances where a

beneficiary has an early stage of ESRD that does not require dialysis, or a beneficiary is

receiving acute dialysis for a limited timeframe. To consistently apply ESRD risk scores to

beneficiaries receiving ongoing dialysis, CMS uses notification from a dialysis facility that the

beneficiary is receiving dialysis in a Medicare certified facility. The dialysis facility submits the

notification to CMS on the CMS-2728 form and the payment system uses this information to

apply an ESRD dialysis risk score. This allows CMS to calculate risk scores that are appropriate

for the dialysis population.

116

Section L. Frailty Adjustment for Fully Integrated Dual Eligible Special Needs Plans

(FIDE SNPs) and PACE Organizations

Frailty for FIDE SNPs

Comment: The majority of commenters who made a remark about frailty for FIDE SNPs

expressed concern about the decrease in the frailty factors and the impact on enrollee frailty

scores, especially in combination with the impact of the proposed CMS-HCC risk adjustment

model. Plans cited a number of reasons for their concern, including: general concerns about the

decline in frailty scores and disproportionate impacts on vulnerable and high-need populations

because of payment decreases; concerns that it will be harder for FIDE SNPs to qualify for

frailty because, in order to qualify, the average level of frailty in the FIDE SNP has to be similar

to that in the PACE program; and concerns that the decreases in frailty factors may be anomalous

for full-benefit and/or partial-benefit dually eligible individuals.

Response: We appreciate commenters’ concern regarding the change in the frailty factors. For

CY 2024, CMS proposed to update the frailty factors used to calculate frailty scores for

beneficiaries enrolled in FIDE SNPs in order to align with the 2024 CMS-HCC risk adjustment

model and, in addition, applied a new technical adjustment to account for potential non-response

bias. In reviewing the initial frailty factors produced by the survey results using CMS’ long-

standing methodology, we noticed differences in some of the underlying patterns of frailty

factors relative to prior years. For example, the full-Medicaid 0 Activities of Daily Living (ADL)

group that typically does not predict additional costs due to limitations of ADLs (i.e., predicted a

negative factor) has a positive factor under the recalibrated frailty factors. CMS believes the

differences in patterns of the factors may result from a number of drivers, including declines in

response rates and the underlying FFS population changes. To address these concerns, CMS

newly applied the Consumer Assessment of Healthcare Providers and Systems (CAHPS)

survey weight designed to adjust for the potential impact of non-response bias in the CAHPS

survey. While the application of the CAHPS survey weight maintains frailty factor patterns that

are largely consistent with prior years, the application of the weight puts downward pressure on a

number of the frailty factors. We do not believe the frailty factors without the CAHPS survey

weight applied are erroneous as they reflect the results of predicted residual costs from the model

for the survey population.

In consideration of the comments received, for CY 2024, CMS is finalizing updated frailty

factors that align with the 2024 CMS-HCC model, but that were not calculated using the CAHPS

survey weight that was proposed in the CY 2024 Advance Notice.

The final frailty factors for CY 2024 can be found in Table III-4. While we are not applying the

CAHPS survey weight for CY 2024, we will continue to evaluate the underlying patterns driving

Consumer Assessment of Healthcare Providers and Systems (CAHPS) surveys

117

the changes in the CY 2024 frailty factors and the appropriateness of applying CAHPS survey

weighting to the frailty factors. CMS will take our findings under consideration when making

future updates to the frailty factors.

We note that we must implement frailty factors that align with the CMS-HCC risk adjustment

model to be used in payment, since the frailty factors are calculated by predicting costs that are

not captured by the CMS-HCC risk adjustment model used for payment. For this reason, we are

unable to delay use of the recalibrated frailty factors, as some commenters requested. However,

the frailty factors being finalized for CY 2024 will be phased in on the same schedule as the Part

C model. Frailty scores for FIDE SNPs in CY 2024 will be calculated by blending 67% of the

frailty scores calculated using the frailty factors associated with the 2020 CMS-HCC risk

adjustment model (finalized in the 2022 Rate Announcement

) and 33% of the frailty scores

calculated using the frailty factors associated with the 2024 CMS-HCC risk adjustment model.

The blended frailty score will be compared to the PACE level of frailty calculated in the same

manner to determine whether that FIDE SNP has a similar average level of frailty as PACE.

Specifically, we note that the PACE minimum is calculated using the same frailty factors as

those used to calculate the frailty scores for FIDE SNPs, so changes in FIDE SNP frailty scores

will be accompanied by a change in how the PACE minimum is calculated. See the 2012 Rate

Announcement for a discussion of how CMS calculates the PACE minimum.

Table III-4 Final Frailty Factors Associated with the 2024 CMS-HCC Model – FIDE SNPs

Number of ADLs

Non-Medicaid

Partial Medicaid

Full Medicaid

-0.066

-0.070

0.158

1-2

0.103

0.203

0.230

3-4

0.201

0.203

0.230

5-6

0.201

0.217

0.248

Comment: Some commenters included a recommendation that CMS provide additional

information about how the change in frailty factors was formulated through rulemaking. A few

commenters recommended delaying the implementation of the CY 2024 CMS-HCC risk

adjustment model with the frailty adjustment, or implementing a phased-in approach. A few

commenters asked for more information or wanted CMS to be more transparent about the

process. A commenter recommended that stakeholders have more time before changes are made

to the frailty adjustment and recommended convening a technical expert panel.

Response: Section 1853(a)(1)(B)(iv) of the Act authorizes an additional payment adjustment that

takes into account the frailty of beneficiaries enrolled in Fully Integrated Dual Eligible Special

Refer to Section K. Frailty Adjustment for PACE Organizations and FIDE SNPs of the 2022 Advance Notice for the frailty

factors associated with the 2020 CMS-HCC risk adjustment model.

Refer to Section L. Frailty Adjustment of the 2012 Rate Announcement for information regarding the establishment of the

PACE minimum.

118

Needs Plans (FIDE SNPs), if the average level of frailty in the FIDE SNP is similar to that in the

PACE program, and CMS has chosen to apply such an adjustment. The frailty adjustment for

FIDE SNPs is part of the risk adjustment methodology. Because section 1853(b) of the Act

provides for the update of risk and other factors used to adjust capitation payments through the

Advance Notice and Rate Announcement, frailty adjustment factors have consistently been

proposed and finalized through the Advance Notice and Rate Announcement since we first

started making these adjustments for frailty for qualifying FIDE SNPs in 2012. As the frailty

factors must be estimated in alignment with the CMS-HCC risk adjustment model being used for

the payment year, it is a component of the methodology used to adjust payments to plans and,

therefore, is appropriately finalized using the process specified in section 1853(b) of the Act. As

previously stated, we must implement frailty factors that align with the CMS-HCC risk

adjustment model to be used in payment, since the frailty factors are calculated by predicting

costs that are not captured by the CMS-HCC model. For this reason, we are unable to delay the

use of the recalibrated frailty factors, as some commenters requested. As discussed previously,

the Advance Notice and Rate Announcement are the appropriate means by which changes are

made to MA payments. This also pertains to frailty factors for FIDE SNPs.

Comment: A few plan commenters recommended allowing additional plan types (e.g., HIDE

SNPs with exclusively aligned enrollment, ESRD C-SNPs, and/or plans that serve any dually

eligible beneficiaries) to be eligible for the frailty adjustment.

Response: By law, CMS must use the same payment methodology for all enrollees in MA plans,

including Special Needs Plans (SNPs), except as explicitly provided for in statute. Section

1853(a)(1)(B)(iv) of the Act authorizes CMS to make frailty-adjusted payments only to certain

dual SNPs – those with fully integrated, capitated contracts with states for Medicaid benefits,

including long term care, and which have similar average levels of frailty as the PACE program.

Thus, CMS cannot make frailty payments to any SNP that does not meet these criteria without

implementing frailty payments program-wide.

CMS has explored ways of incorporating frailty into the risk adjustment model in order to

account for frailty when making risk adjusted payments to all plans and found challenges with a

number of approaches (see the “Evaluation of the CMS-HCC Risk Adjustment Model,”

published March 2011).

In addition, as directed by section 17006(f)(2)(b) of the 21st Century

Cures Act, the Government Accountability Office issued a report on issues related to

incorporating functional status into MA risk adjustment in 2018.

This study found a number of

challenges with incorporating frailty into the model, including that “stakeholders could face

substantial challenges if the risk adjustment model were revised to account for beneficiary

Pope, Gregory C.; Kautter, John; Ingber, Melvin J.; Freeman, Sara; Sekar, Rishi; and Newhart, Cordon. (March 2011).

Evaluation of the CMS-HCC Risk Adjustment Model.

GAO Medicare Advantage Benefits and Challenges of Payment Adjustments Based on Beneficiaries’ Ability to Perform Daily

Tasks.

119

functional status, in part because this information is not readily available.” The CMS-HCC risk

adjustment model uses demographic factors and diagnoses to predict relative costs for

subpopulations, with the frailty adjustment used to predict expenditures for community

beneficiaries with functional impairments that are unexplained by the risk adjustment model

alone. Because the frailty factors are calculated using the residual of the CMS-HCC risk

adjustment model (the difference between the predicted expenditure amounts and the actual

expenditure amounts), and frailty scores have an average value of zero, the application of a

frailty adjustment to all MA plans would result in many plans receiving a negative frailty

adjustment.

Frailty for PACE Organizations

Comment: Some commenters expressed concerns with using the HOS-M survey to estimate

frailty because of low response rates, and that reliance on the HOS-M for frailty adjustment does

not consider the challenges faced by people with dementia in completing the survey. Most of

these commenters urged that, if dementia could not be included in the risk adjustment model

used to pay PACE organizations for CY 2024, that CMS modify the CY 2023 HOS-M survey

administration protocol to allow PACE organizations to proactively offer completion assistance

for the survey to their participants living with dementia to increase the likelihood that they are

adequately represented in the survey’s results.

Response: CMS estimates frailty factors to explain additional costs not explained by diagnoses in

the CMS-HCC risk adjustment model. To the extent that these costs are not predicted by the

model, they are likely to be reflected in the frailty factors. CMS calibrates the frailty factors by

regressing the residual, or unexplained costs, from the CMS-HCC risk adjustment model on

counts of activities of daily living (ADLs). Although total costs are included in the calibration of

the 2017 CMS-HCC risk adjustment model, and the associated frailty factors help predict overall

costs where diagnoses are not fully predictive, results for individual organizations may differ due

to differences between the sample used for model calibration and the populations enrolled in

individual plans.

CMS acknowledges the concerns related to improving the response rates for the HOS-M for

PACE participants generally, and among those with dementia. The responses from this survey

are used to determine a beneficiary’s limitations in ADLs for the calculation of a frailty score.

For the HOS-M, CMS will continue to allow a beneficiary to designate a proxy for responding to

the survey, including a PACE staff member. In addition, PACE organizations are able to provide

additional beneficiary contact information to support response rates. We collect survey data in

this manner so that there is consistency across PACE organizations, as this helps to ensure

equitable frailty results for payment.

120

Section M. Medicare Advantage Coding Pattern Adjustment

Comment: Many commenters opposed CMS’ proposed 5.9 percent 2024 coding pattern

adjustment and provided alternate alternative recommendations to the statutory minimum coding

pattern adjustment of 5.9 percent, as summarized below:

• Higher adjustment factor: Several commenters recommended a higher adjustment factor

than the statutory minimum, which they state is inadequate to adjust for differential

patterns of coding between MA and FFS. Commenters expressed concern that the

statutory minimum does not account for the full impact of coding pattern differences, and

multiple commenters highlighted analyses from MedPAC that the coding adjustment

factor should be several percentage points higher. These commenters stated their belief

that excess spending is accelerating the depletion of the Medicare Trust Funds and the

potential savings from fully accounting for the coding pattern differential would increase

solvency of the Trust Funds. A few commenters that recommended a higher coding

pattern adjustment expressed concern that the current application of the minimum

adjustment and the risk adjustment model incentivize plan sponsors to code their

enrollees with as many conditions as possible, driving up payment rates. One commenter

expressed concern that CMS’ current methodology does not address the underlying

causes of coding intensity, thereby undermining the goal of plans competing on the basis

of quality and costs. Another commenter noted their belief that increased payments to

MA plans do not result in better care.

• Lower adjustment factor: One commenter stated that the coding pattern adjustment

results in inappropriately low risk scores that negatively impacts payment for many

nonprofit regional health plans.

• Specific Methodological Recommendations:

o Demographic Estimate of Coding Intensity (DECI). A few commenters

recommended the incorporation of the DECI method to calculate a coding pattern

adjustment factor. Under the assumption that MA does not receive adverse or

favorable selection relative to FFS in terms of health status, the recommended

DECI method controls for demographics, estimating the coding pattern

adjustment by comparing MA risk relative to FFS risk using the CMS-HCC risk

adjustment model, and comparing that relationship against MA risk versus FFS

risk using the Adjusted Average Per Capita Cost (AAPCC) model that is based on

demographics only and was used in payment prior to 2000.

o Targeted approaches:

▪ General targeted comments. Several commenters expressed concern that coding

patterns across the MA landscape are heterogeneous and that failure to recognize

121

these differences across plans by applying an across-the-board coding pattern

adjustment could result in an inequitable outcome. A few commenters

recommended targeted approaches, because of their concern that certain MA

organizations code much more aggressively than others with higher levels of

coding intensity due to various structural payment incentives, including payments

between MA organizations and their contracted providers. Other commenters

stated their concern about the current application of the factor because it does not

adequately adjust for risk score increases above the average, and disadvantages

plans serving primarily low-income and historically underserved communities

that have less administrative resources to focus on diagnosis coding.

▪ Segmented approach. A few commenters suggested a segmented approach to

coding pattern adjustments that recognizes different levels of coding patterns

among plans, such that the lowest coding factor is applied to lower coding plans

while the highest factor is applied to higher coding plans.

▪ Contract-specific approach. A few commenters recommended tailoring the MA

coding pattern adjustment to the relative level of coding intensity seen in

individual MA contracts – rather than the across-the-board coding pattern

adjustment that CMS applies today to all MA contracts. A few commenters

believe that CMS should consider increasing the MA coding pattern adjustment

for all contracts and consider using its statutory authority to vary the coding

pattern adjustment by contract.

A few commenters had recommendations to calibrate the model using different data to address

coding pattern differences between MA and FFS. One commenter recommended a multipronged

approach to addressing coding pattern differences in MA and FFS. Their recommendation

included three parts: 1) develop a risk adjustment model that uses two years of FFS and MA

diagnostic data; 2) exclude diagnoses that are documented only on health risk assessments from

either FFS or MA; and then 3) apply a coding adjustment that fully accounts for the remaining

differences in coding between FFS Medicare and MA plans.

Response: On an annual basis, CMS analyzes coding pattern differences and determines what the

coding pattern adjustment factor should be. Section 1853(a)(1)(C)(ii) of the Act establishes a

minimum MA coding pattern adjustment, which was originally adopted beginning with 2014

payment. The current statutory minimum coding pattern adjustment is 5.9%. We have found that

the minimum adjustment is sufficient to reflect differences in coding patterns between MA plans

and providers under FFS Parts A and B. CMS continues to believe that applying a uniform

adjustment is an appropriate approach. Therefore, we are finalizing our proposed MA coding

pattern adjustment factor for CY 2024.

122

We appreciate the extensive and thoughtful comments and feedback we received on this

proposal. Ensuring that the coding pattern adjustment policy appropriately addresses differences

in coding patterns between the FFS program and MA is essential and we will consider these

recommendations in the development of future coding pattern adjustment proposals.

Comment: A few commenters supported CMS’ proposed 5.9 percent 2024 coding pattern

adjustment.

Response: CMS appreciates the support of the commenters. CMS is finalizing the proposed

adjustment of 5.9 percent for CY 2024.

Comment: One commenter requested sufficient time and information to comment on any

potential changes to the MA coding pattern adjustment in the future.

Response: CMS appreciates the comment. Section 1853(b)(2) requires that CMS provide notice

of proposed changes in the methodology and assumptions for setting MA capitation rates and

risk and other factors used to adjust the capitation payments, with a comment period of at least

30 days to comment on the proposed changes. We will continue to consider additional ways in

which we can engage with stakeholders should we consider changes to the MA coding pattern

adjustment.

Comment: A few commenters recommended that CMS move to a risk adjustment model based

on MA encounter data to improve payment accuracy, which would also eliminate the need for a

coding pattern adjustment.

Response: We appreciate the recommendation to move to an encounter-data based model. CMS

recognizes the benefits of moving to a payment system that is based on MA experience. As

discussed above, public comment would be an essential part of developing an encounter data-

based model.

Comment: A few commenters believed that it is fundamentally incorrect to assume any observed

coding differentials between the FFS and MA populations are driven by inappropriate coding on

the part of MA plans and requested that CMS recognize that higher coding does not necessarily

equate to wrong coding. One commenter stated that CMS should consider that differences in

coding stem from the fact that FFS is unmanaged and under-coded, and that the differences

actually demonstrate the value of MA plans in diagnosing and appropriately managing members’

conditions. One commenter stated that the MA coding pattern adjustment was redundant to the

many structures in place to reduce coding intensity such as the Risk Adjustment Data Validation

audit (RADV).

Response: The MA coding pattern adjustment is not intended to adjust for inaccurate coding, but

it is intended to account for program-wide differences in coding patterns between MA and FFS.

CMS applies the MA coding pattern adjustment to adjust for the impact on MA risk scores of

123

coding patterns that differ from FFS coding, which is the basis of the CMS-HCC model. More

specifically, Section 1853(a)(1)(C)(ii) of the Social Security Act requires the Secretary to reduce

payments to MA organizations by at least 5.9% to account for “differences in coding patterns

between Medicare Advantage plans and providers under part[s] A and B.” This mandatory

adjustment is applied to each beneficiary’s risk score at every MA plan, which reflects how

section 1853(a)(1)(C)(ii) focuses on coding pattern differences at a general, program-wide level.

Specifically, section (a)(1)(C)(ii) addresses the incentive for all MA plans to identify more valid,

supported codes on all HCCs, and therefore report more diagnoses (which we do not assume to

be inaccurate when evaluating coding pattern differences between FFS and MA). By reducing

each individual risk score by a specified percentage, the coding pattern adjustment corrects for

overall, net differences between MA and FFS coding.

Section N. Normalization Factors

CMS did not receive comments on the methodology proposed to calculate the RxHCC risk

adjustment model normalization factors for CY 2024. CMS is finalizing the RxHCC

normalization factor methodology as proposed.

Comment: For the CMS-HCC and CMS-HCC ESRD risk adjustment models, many commenters

were in support of the continued exclusion of the CY 2021 risk score from the calculation of the

normalization factors for CY 2024, consistent with the policy finalized for CY 2023. Some

commenters stated their agreement with CMS’ belief that the decreased 2021 risk score is driven

by reduced utilization in 2020 due to the pandemic and, therefore, should be excluded from the

normalization factor. Several commenters were, in general, supportive of the methodology

proposed to calculate the normalization factors for CY 2024, with a couple expressing their

support only for the proposed methodology for the newer models (the CMS-HCC model

proposed in the 2024 Advance Notice and the CMS-HCC ESRD models finalized in CY 2023).

Response: CMS appreciates the support of the commenters. We are finalizing the normalization

factor methodology for the CMS-HCC and CMS-HCC ESRD risk adjustment models as

proposed.

Comment: Several commenters were generally concerned that the proposed normalization

factors for the CMS-HCC and CMS-HCC ESRD risk adjustment models are overstated, with one

commenter stating their belief that this will translate to fewer benefits for beneficiaries. One

commenter had concerns about applying a common normalization factor across model segments

because they believe that the impact of the COVID-19 pandemic was not the same across

segments, potentially leading to disproportionate impacts of the normalization factor across plans

depending on their beneficiary population. A couple of commenters suggested a more nuanced

approach be used for the inclusion of the 2022 risk score in the normalization factors for the

newer models.

124

A few commenters were specifically concerned about CMS’ proposed methodology to exclude

the 2021 risk score (based on 2020 dates of service) from the calculation of the risk adjustment

model normalization factors for CY 2024. A couple of these commenters believed that excluding

the 2021 risk score data continues to deviate from CMS’ longstanding methodology to use risk

scores based on the most recent data available so as to reflect recent FFS experience. There was

concern by a couple of commenters that excluding more recent years (for the older models), or

weighting post-pandemic risk scores equally with pre-pandemic risk scores (for the newer

models), may not be appropriate as the risk score experience may not be the same in the two time

periods. A couple of commenters also stated that removing the 2021 risk score data from the

normalization methodology increases the negative normalization adjustment, resulting in further

reduction to risk scores and resulting payments. One of these commenters believed CMS’

proposed methodology was contrary to CMS’ data that showed continued lower risk scores in

FFS while another questioned the accuracy of the normalization factors when excluding the 2021

risk score. One commenter requested that CMS explain how it plans to address the anomalous

2020 utilization patterns in future years, and voiced concern that utilization in 2021 and 2022

may also be anomalous. This commenter asked that CMS be consistent with how it uses the 2021

risk score in normalization in that, if it is excluded for CY 2024, that it continues to be excluded

in the future.

A couple of commenters supported CMS’ proposed methodology to calculate the normalization

factors for the newer models (those with a 2019 or 2020 denominator) using risk scores from

2018-2022, excluding 2021, but opposed the proposed methodology for the older models (those

with a 2015 denominator) to exclude both 2021 and 2022 risk scores. These commenters

believed that the CMS should be consistent with its methodology across models and that the

2022 risk score should be included for both newer and older models so as not to overstate the

normalization factor for the older models, which would lead to insufficient MA payments. They

believe that the pre-pandemic experience may not reflect post-pandemic experience and by

excluding more recent years from the older models, CMS is further deviating from its

longstanding methodology to use recent FFS experience in the calculation of the normalization

factor. Alternate methodologies for including 2022 in the normalization factor for the older

models were suggested by a few commenters, such as using more years in the trend, projecting

the trend differently, weighting 2022 risk scores more heavily to reflect more relevant and recent

FFS experience, or blending.

Several commenters also requested that CMS be more transparent in its normalization factor

calculation by releasing more data, providing more methodological information, further

explaining its rationale for the proposed policy, or establishing a Technical Expert Panel to

discuss the calculation of the normalization factor.

Response: CMS appreciates commenters’ concerns regarding the calculation of the

normalization factors for CY 2024. We believe, however, the proposed methodologies – using a

linear approach with the most recent five years of data, with the exclusion of 2021 (2018–2022,

125

excluding 2021) for newer models, and using a linear approach with the exclusion of 2021 and

2022 (2016-2020) for the older models – are better projections of the applicable average FFS risk

score in 2024.

The goal of the normalization factor is to accurately predict the FFS risk score in the payment

year, thereby maintaining an average FFS risk score of 1.0 across the entire FFS population.

CMS believes that the inclusion of the 2021 risk score in the slope calculation will result in a

projected risk score (i.e., normalization factor) that is significantly below what the actual average

FFS risk score is likely to be in 2024 whether the risk score is based on the current or proposed

CMS-HCC model. In other words, including the 2021 risk score in either projection results in an

unrealistic estimate of the 2024 FFS risk score.

Including the 2021 risk score and applying our typical methodology for the current CMS-HCC

model yields a CY 2024 normalization factor that is lower than the actual 2022 FFS risk score.

For the proposed model, if the 2021 risk score were included in the calculation using our typical

methodology, the CY 2024 normalization factor would be 0.8% higher than the 2022 risk score.

CMS believes it is unlikely that 2024 risk scores will be lower than 2022 risk scores, or that risk

scores will grow by only 0.8% over two years. In addition, in all of the years used to identify the

trend in risk scores prior to 2021, risk scores progressively increased; the decreases in utilization

in 2020 were irregular due to the pandemic. The objective of the normalization factor is to

project the payment year risk scores as accurately as possible to maintain the 1.0, given the

information known at the time the projected scores are calculated. Given this objective, CMS

believes that the decreases in utilization in 2020 due to the pandemic are not reflective of future

health care utilization and should not be included in the calculation of the normalization factors.

Including the 2022 risk score and applying our typical methodology for the current CMS-HCC

model yields a CY 2024 normalization factor that is lower than the 2022 FFS risk score. As

noted above, CMS believes it to be unlikely that 2024 risk scores will be lower than 2022 risk

scores. Given the increase in the 2022 actual FFS risk score relative to 2021 and the continuous

increase in the average FFS risk score prior to the pandemic, it is not reasonable to apply a

normalization factor that is lower than the most recent risk score data point in the trend. While

we think it is important to incorporate more recent years of data in the trend to reflect current

risk, updating the data must be balanced with projecting a risk score that is reflective of what the

average 2024 FFS risk score is likely to be in order to establish the appropriate normalization

factor.

CMS appreciates commenters’ concerns about the impact of the COVID-19 pandemic on

utilization and diagnoses submission, and the potential effects on risk adjusted payments. CMS

carefully considered the use of the 2021 FFS risk score in the calculation of the slope used to

project the normalization factors for the CMS-HCC risk adjustment models for CY 2024. The

2021 risk score, which is based on diagnoses from 2020 dates of service, is significantly lower

than the 2020 risk score, which was based on diagnoses from 2019 dates of service. Prior to

126

2021, risk scores progressively increased in the years used to identify the trend in risk scores. We

believe that the decrease in the 2021 risk score is driven primarily by reduced utilization in 2020

due to the pandemic. While CMS understands the uncertainty surrounding the future impact of

the COVID-19 pandemic and the use of the 2020 utilization data, every year CMS re-evaluates

the data and bases policy decisions on the information available. We will continue to monitor

and analyze underlying risk score trends and their drivers. The public will have an opportunity to

comment on future proposed policies.

While there is inherent uncertainty with any prediction of future values, the proposed approaches

to calculating the normalization factors for CY 2024 maintain the stability of using our

longstanding five-year linear slope methodology while balancing the impact of the pandemic on

the normalization factor projection and the progressive increase in risk scores evident in the

historical trend prior to 2021.

Comment: A couple of commenters requested further rationale for CMS’ proposal to include

2020 utilization and data for development of the MA benchmarks and growth rates but not for

the CY 2024 normalization factors. A couple of commenters also questioned CMS’ use of 2020

as the denominator year for the proposed CMS-HCC risk adjustment model because the 2020

denominator uses 2020 expenditures/2020 dates of service. These commenters were concerned

that CMS is being inconsistent in its use of the 2020 experience across the policies proposed in

the 2024 Advance Notice.

Response: We understand that commenters are concerned about the treatment of 2020 data in

some of CMS’ MA payment policies. CMS carefully considered the appropriateness of 2020

data and made a determination based on how the data is being used (e.g., as part of an average

versus part of a trend), and the reasonableness of the impact of the data on what is being

measured. As described in more detail below, the impact of an inconsistent data point differs

when used to calculate an average versus a projected value. Prior to establishing MA

benchmarks for CY 2024, the trends in the 2020 FFS data used to establish the benchmarks were

analyzed. Some specific regions did experience decreased per-capita costs while other regions

experienced increased per-capita costs when compared to the 2019 national average per-capita

costs. However, because the ratebook FFS average geographic adjustments (AGAs) use data to

develop a relative index that averages out to 1.0, the level of the 2020 FFS claims is not

impactful for this measure. Furthermore, for ratebook development, CMS uses an average of five

years of FFS experience for each county, so annual fluctuations and anomalies in the data that

may occur for a variety of reasons are mitigated. Calculating and using a five-year average

provides stability in the rates despite local or regional events, such as natural or weather-related

disasters, and varying impacts from nationwide events, such as pandemics.

Distinct from calculating the benchmarks, normalization factors are calculated using five years

of historical data to create a trend that is projected out to a future payment year. As described in

the CY 2023 Advance Notice, when calculating a trend, one anomalous data point can have a

127

large impact on the projected value, which can pull the slope up or down significantly and lead to

a projection that does not reasonably estimate a future value. This trending issue does not apply

with rebasing where historical data are used to calculate a five-year rolling average in the AGA

calculation for ratebook development, so the impact of any one year of anomalous utilization is

moderated by four other years of data.

Prior to proposing the CY 2024 normalization factors for the CMS-HCC and CMS-HCC ESRD

risk adjustment models, CMS carefully considered the impact of using the 2021 risk score (2020

dates of service) in the calculation of the slope used to project the 2024 FFS risk score (i.e., the

normalization factor). CMS believes that the decrease in utilization due to the pandemic was

irregular and not reflective of future health care utilization, and, if the data were used to project a

future risk score, would result in an underestimate of the normalization factor. The policy of

excluding the 2021 risk score (2020 dates of service) from the normalization factor calculation is

consistent with the approach we used to project FFS USPCCs in the 2023 Rate Announcement in

that 2020 data are excluded. Like the methodology used to calculate normalization factors, the

methodology used to estimate national FFS spending projects a future value based on a trend.

For both estimates, which rely on trending and projecting using historical data, CMS consistently

excluded the 2020 data.

As noted by commenters, CMS used a 2020 denominator to create relative factors for the

proposed CMS-HCC risk adjustment model. In order to use the risk adjustment model to

calculate risk scores for payment, we create relative factors for each demographic factor and

HCC in the model. The relative factors are used to calculate risk scores for individual

beneficiaries. CMS creates relative factors by dividing all the model-based dollar coefficients by

the average per capita predicted expenditures for a specific year (the “denominator year”). To

calculate a model denominator, in this case the average per capita predicted expenditures for

2020 (i.e., the 2020 denominator), we do not use actual expenditures, but rather, we apply the

proposed CMS-HCC model to 2019 diagnoses from a cohort of 2020 beneficiaries, and produce

a predicted average per capita expenditure for 2020. Therefore, actual 2020 experience –

utilization or expenditures – is not reflected in the 2020 denominator.

Section O. Sources of Diagnoses for Risk Score Calculation for CY 2024

CMS did not receive comments within the scope of the policy as to which sources of diagnoses

are used for risk score calculation for CY 2024.

For non-PACE organizations, for CY 2024, CMS will continue the policy adopted in the CY

2023 Rate Announcement to calculate risk scores for payment to MA organizations and certain

demonstrations using only risk adjustment-eligible diagnoses from encounter data and FFS

claims.

For PACE organizations, for CY 2024, CMS will continue using the same method of calculating

risk scores under the CMS-HCC and ESRD models that we have been using since CY 2015,

128

which is to pool risk adjustment-eligible diagnoses from the following sources to calculate a

single risk score (with no weighting): (1) encounter data, (2) Risk Adjustment Processing System

(RAPS) data, and (3) FFS claims.

Attachment IV. Responses to Public Comments on Part D Payment Policy

Section A. RxHCC Risk Adjustment Model

Comment: Of the commenters that commented specifically on the proposal to continue using the

2023 RxHCC model for CY 2024, most expressed concerns about the impact of using the 2023

model in CY 2024 given the plan design changes called for in the IRA. These commenters

believed that changes such as the elimination of cost-sharing in the catastrophic phase of

coverage and cost-sharing caps on insulin and vaccines would result in increased plan liability

that would not be accounted for in the 2023 RxHCC model. Some commenters expressed

additional concerns about the impact that underestimation of plan liability would have on low-

income beneficiaries, beneficiaries with high-cost medications, and beneficiaries in D-SNP

plans. Some commenters supported updating the RxHCC model in CY 2025 to reflect plan

design changes to appropriately reflect relative risks of Part D enrollees. A couple of

commenters recommended CMS take specific steps in CY 2024 to account for the proposal to

use the 2023 RxHCC model in CY 2024 (e.g., providing flexibility to adjust margins during the

rebate reallocation process).

Response: CMS appreciates the commenters’ concerns. As noted in the CY 2024 Advance

Notice, the IRA was enacted well into the timeline needed to conduct a revision to the RxHCC

risk adjustment model to reflect these recent benefit changes for CY 2024. We acknowledge that

these benefit changes can result in changes to plan liability for beneficiaries with conditions that

result in an increased likelihood of reaching the catastrophic phase of the benefit. It requires

extensive time to prepare the data to update the model, run the recalibration and calculate the

relative factors, review the output, and finalize an updated model for publication in the Advance

Notice. For example, for the RxHCC risk adjustment model, this work includes re-mapping all

the Prescription Drug Event data to reflect the new plan liability, and re-estimating the RxHCC

coefficients based on the updated plan liability. For these reasons, the insufficient time CMS had

to recalibrate the RxHCC risk adjustment model to account for plan design changes called for by

the IRA made an update infeasible.

We note that any changes to plan liability as a result of IRA changes to the Part D benefit are not

expected to be large. Further, to the extent that any plans’ resulting revenue need for the plan’s

liability under the CY 2024 benefit sufficiently exceeds what they projected in their bid, then the

Part D sponsor would be compensated for some of their losses through the Part D risk corridors.

As discussed in the Advance Notice, we will be recalibrating the model based on the updated

benefit structure for CY 2025.

129

For CY 2024 CMS will continue to use the RxHCC Risk Adjustment Models as discussed in the

2024 Advance Notice:

• For Non-PACE Organizations: For CY 2024, we will continue to use the 2023 RxHCC

risk adjustment model to adjust direct subsidy payments for Part D benefits offered by

stand-alone Prescription Drug Plans (PDPs) and Medicare Advantage-Prescription

Drug Plans (MA-PDs).

• For PACE Organizations: For CY 2024, CMS will continue to use the 2020 RxHCC

risk adjustment model to calculate Part D risk scores.

Comment: A few commenters expressed concerns about the combined effects of changes to the

CMS-HCC risk adjustment model and the RxHCC model. One commenter believed that the

CMS-HCC changes could threaten the ability of MA plans to use rebate dollars to buy down the

Part D premiums, while another believed that the removal of codes from the CMS-HCC model

would cause discrepancies between plan coding activities for risk adjustment between the Part C

and Part D programs. Another commenter suggested revising RxHCC model coefficients for any

conditions, such as diabetes, which were being removed or disincentivized in the CMS-HCC

model.

Response: CMS appreciates the commenters’ concerns. These two models have always had key

differences, among them that the CMS-HCC risk adjustment model predicts the costs of Part A

and Part B benefits, while the RxHCC model predicts plan liability for prescription drugs

covered under the Part D program. Because the models predict different costs, it has always been

the case that they have HCCs that are formulated differently or that are in one model and not the

other. The CMS-HCC risk adjustment model is being updated to improve the prediction of costs

for the Part A and Part B benefits, using updated data and a reclassification of the HCCs so they

are based on the ICD-10 classification system. For CY 2023, the RxHCC model was updated to

improve the prediction of plan liability for Part D prescription drug costs by calibrating the

model on more recent data years and using newly built HCCs reflected a reclassification using

ICD-10 codes. In this way, for CY 2024, the RxHCC and CMS-HCC models will be more

closely aligned. We do not anticipate that changes to the CMS-HCC model will cause

discrepancies in plan coding activities between the Part C and Part D programs.

Comment: Several commenters recommended that CMS examine other methods where the

underlying data and structure of the RxHCC model could be modified for CY 2025. These

commenters believed that the model underpredicts plan liability for common therapeutic areas,

especially for newer, high-cost treatments and for low-income beneficiaries. Some of these

commenters also believed that the redesign of the Part D benefit in CY 2025 could lead to

incentives for plans to select beneficiaries with more favorable risk, so the model should be

updated to mitigate this potential.

130

Response: CMS appreciates the commenters’ concerns. As discussed in the Advance Notice, we

will be recalibrating the model based on the updated benefit structure for CY 2025 and will take

this feedback into account as part of the development and recalibration process. We consistently

look for ways to improve our models and thank the commenters for their thoughts.

Comment: A few commenters suggested steps such as including drug utilization markers similar

to those in the commercial market, as well as updating the model with more recent claims data to

reflect more recent drug costs and improve the predictive power of the model.

Response: CMS thanks the commenters for their suggestions. We understand that in certain

programs such as the Marketplace where a combined medical and drug model is utilized, the

methodological approach for predicting relative costs that are specific to that population may be

different from the approach used in MA. CMS uses the RxHCC risk adjustment model to adjust

the direct subsidy payments for Part D benefits offered by stand-alone prescription drug plans

and MA-Part D plans. Having the RxHCC model used to predict drug costs separate from the

CMS-HCC model used to predict medical costs enables a single model – the RxHCC model – to

account for differences in predicted plan liability for prescription drugs among distinct subgroups

of Part D eligible beneficiaries.

Comment: A majority of commenters recommended that CMS release information about the

recalibration of the RxHCC model that incorporates the updated benefit structure in advance of

the CY 2025 Advance Notice, with many of these commenters also recommending that CMS

allow for a 60-day comment period for the RxHCC model so that plans have more time to

evaluate the methodological changes.

Response: We thank the commenters for their recommendations.

Section B. Sources of Diagnoses for Part D Risk Score Calculation for CY 2024

Please refer to Attachment III, Section O. for comments and responses regarding sources of

diagnoses.

Section C. Inflation Reduction Act of 2022 Part D Benefit Design Changes

Comment: Many commenters expressed support for CMS’ implementation of the IRA changes

in place for 2024.

Response: CMS thanks the commenters for their support.

Comment: A commenter expressed opposition to CMS’ implementation of the IRA changes in

place for 2024, stating that the changes will lead to reduced benefits, increased cost-sharing, and

increased premiums.

Response: CMS appreciates the commenter’s concerns. We note that the IRA changes in place

for 2024 are mandated by statute and anticipate that these changes will increase the affordability

131

and accessibility of Part D drugs for Medicare beneficiaries. Furthermore, the premium

stabilization requirements will prevent unaffordable increases in premiums owed by Part D

beneficiaries.

Comment: A commenter requested clarification that the low-income cost sharing subsidy will

continue to cover the full cost of a claim with respect to ACIP-recommended vaccines and

covered insulin products for LIS beneficiaries up to the low-income patient payment in the

coverage gap. Another commenter argued that the low-income cost sharing subsidy used to

calculate the coverage gap discount with respect to covered insulin products for LIS beneficiaries

should be calculated based on the full cost of the drug rather than the maximum LIS copayment.

Response: For LIS beneficiaries in the coverage gap in 2024, the low-income cost-sharing

subsidy will continue to cover the full cost of an ACIP-recommended vaccine and a covered

insulin product up to the maximum LIS copayment in accordance with section 1860D-

14(a)(1)(C) of the Act. As such, the low-income cost sharing subsidy will continue to cover the

full cost of ACIP-recommended vaccines and covered insulin products up to the lesser of the

applicable copayment amount or nominal copayment amount. LIS-eligible beneficiaries are not

applicable beneficiaries and do not receive a coverage gap discount. Additionally, we clarify that

when we stated in the CY 2024 Advance Notice that no cost will be incurred by LIS and non-LIS

beneficiaries for ACIP-recommended adult vaccines and, therefore, no costs will count as

TrOOP toward the OOP threshold for beneficiary progression into the catastrophic phase, we

were speaking only of the amount actually paid by the beneficiary and not the low-income cost

sharing subsidy amount.

Section D. Medicare Part D Benefit Parameters: Annual Adjustments for Defined

Standard Benefit in 2024

Comment: A few commenters requested that CMS consider the impact of the 2023 pharmacy

price concession rule when calculating the API and CPI in the Rate Announcement. A

commenter believes that the pharmacy price concession rule will reduce the negotiated price of

drugs and patient drug spending, and therefore the effect of the pharmacy price concession rule

should be considered when calculating the API and CPI. Some commenters also requested that

CMS consider the changes to the coverage of ACIP-recommended vaccines and covered insulin

products in the calculation of the API and CPI for 2024.

Response: CMS appreciates the commenters’ suggestions and concerns. Under section 1860D-

2(b)(6) of the Act, the API is defined as the annual percentage increase in average per capita

aggregate expenditures for covered Part D drugs in the United States for Part D eligible

individuals for the 12-month period ending in July 2023. The calculation does not involve a

projection of the time period where the pharmacy price concession rule is effective. Additionally,

changes to the coverage of ACIP-recommended vaccines and covered insulin products will not

impact the calculation of the API for 2024 because the calculation is based on total drug

132

spending. Our estimate of API includes expected increases in the total drug spending as a result

of additional coverage required for ACIP-recommended vaccines and covered insulin products

during January through July 2023.

Section 1860D-14(a)(4) of the Act requires CMS to use the annual percentage increase in the

CPI for the 12-month period ending in September of the previous year. The calculation also does

not involve a projection of the time period where the pharmacy price concession rule is effective.

Finally, the annual percentage increase in the CPI is determined using the actual September 2022

value as calculated by the Bureau of Labor Statistics and an estimate of the September 2023 CPI

based on projections from the President’s FY2024 Budget.

Comment: A commenter also requested more transparency as to how the value of the 1.09

adjustment that CMS made to the calculation of total gross covered drug costs for 2024 was

determined.

Response: We evaluated the impact of the change in coverage for ACIP-recommended vaccines

and covered insulin products according to Part D historic claims data in 2021 by using

continuance tables for the LIS and Non-LIS populations, and calculating the allowed cost

corresponding to the true out-of-pocket cost (TrOOP) threshold. The adjustments to the

calculation of the allowed cost at the catastrophic threshold are necessary because beneficiaries

take a longer time to reach the TrOOP threshold, and, thus, the catastrophic phase when they pay

less cost sharing prior to the catastrophic phase.

These adjustments have also been updated from those in the CY 2024 Advance Notice to reflect

that the low-income cost sharing subsidy, when applicable, will continue to count toward

TrOOP. As noted above, we clarify that when we stated in the CY 2024 Advance Notice that no

cost will be incurred by LIS and non-LIS beneficiaries for ACIP-recommended adult vaccines

and, therefore, no costs will count as TrOOP toward the OOP threshold for beneficiary

progression into the catastrophic phase, we were speaking only of the amount actually paid by

the beneficiary and not the low-income cost sharing subsidy amount.

Section E. Part D Calendar Year Employer Group Waiver Plans Prospective Reinsurance

Amount

Comment: One commenter supported our policy of paying prospective reinsurance amounts to

Part D EGWPs and recommended that CMS add a trend adjustment to the methodology so that

prospective reinsurance payments take into account the amount by which reinsurance is

projected to increase in the current payment year relative to the most recently reconciled

payment year.

Response: CMS thanks the commenter for their support and recommendation. We do not believe

it would be appropriate to adjust prospective reinsurance payments for CY 2024 by a trend factor

133

when we did not propose to do so in the Advance Notice. Although we decline to add a trend

factor at this time, we will consider this recommendation as we continue to refine our

methodology for future years

Section F. Part D Risk Sharing

Comment: A commenter suggested that CMS consider narrowing risk corridors to help manage

premium increases on beneficiaries.

Response: We appreciate the commenter’s suggestion. Under section 1860D-15(e)(3)(C) of the

Act and § 423.336(a)(2)(ii), CMS may establish a risk corridor with higher threshold risk

percentages for Part D risk sharing. However, the statute does not permit CMS to narrow the

corridors relative to the CY 2011 thresholds.

134

Attachment V. Final Updated Part D Benefit Parameters for Defined Standard Benefit,

Low-Income Subsidy, and Retiree Drug Subsidy

Table V-1. Updated API and CPI for 2024

Annual percentage

trend for 2023

Prior year

revisions

API for 2024

API

6.42%

1.50%

8.01%

September CPI (all items, U.S. city average)

3.81%

3.87%

7.83%

Table V-2. Updated Part D Benefit Parameters for Defined Standard Benefit, Low-Income

Subsidy and Retiree Drug Subsidy

2023

2024

Standard Benefit

Deductible

$505

$545

Initial Coverage Limit

$4,660

$5,030

Out-of-Pocket Threshold

$7,400

$8,000

Total Covered Part D Spending a t Out-of-Pocket Threshold for Non-

Applicable Beneficiaries (1)(6)

$10,516.25

$11,477.39

Estimated Total Covered Part D Spending for Applicable Beneficiaries

(2)(6)

$11,206.28

$12,447.11

Minimum Cost Sharing in Catastrophic Coverage Portion of the Benefit

Generic/Preferred Multi-Source Drug

$4.15

Not Applicable

Other

$10.35

Not Applicable

Full Subsidy-Full Benefit Dual Eligible (FBDE) Beneficiaries (3)

Deductible

$0.00

Copayments for Institutionalized Beneficiaries [category code 3]

Copayments for Beneficiaries Receiving Home and Community-Based

Services] [category code 3] (4)

$0.00

Maximum Copayments for Non-Institutionalized Beneficiaries

Up to or at 100% FPL [category code 2]

Up to Out-of-Pocket Threshold

$0.00

Generic/Preferred Multi-Source Drug

$1.45

$1.55

Other

$4.30

$4.60

Above Out-of-Pocket Threshold

Between 100% and 150% of FPL [category code 1]

Up to Out-of-Pocket Threshold

$0.00

Not Applicable

Generic/Preferred Multi-Source Drug

$4.15

$4.50

Other $10.35 $11.20

Above Out-of-Pocket Threshold

$0.00

Not Applicable

These parameters reflect additional plan coverage required for covered insulin products under section 1860D-2(b)(9) of the

Act, as added by section 11406 of the IRA, and ACIP-recommended vaccines under section 1860D-2(b)(8) of the Act, as added

by section 11401 of the IRA.

135

2023

2024

Full Subsidy-Non-FBDE Beneficiaries (3)

Applied or eligible for QMB/SLMB/QI or SSI, income at or below 135%

FPL in 2023 or at or below 150 % FPL for 2024 and beyond and

resources ≤ $9,090 (individuals, 2023) or ≤ $13,630 (couples, 2023)

[category code 1] (5)

Deductible

$0.00

Maximum Copayments up to Out-of-Pocket Threshold

Generic/Preferred Multi-Source Drug

$4.15

$4.50

Other

$10.35

$11.20

Maximum Copayments above Out-of-Pocket Threshold

$0.00

Not

Applicable

Partial Subsidy (3) (Revised to Full Subsidy Effective 1/1/24)

Applied and income below 150% FPL and resources below $15,160

(individual, 2023) or $30,240 (couples, 2023) [category code 4] 5

Deductible

$104

Not Applicable

Coinsurance up to Out-of-Pocket Threshold

15%

Not Applicable

Maximum Copayments above Out-of-Pocket Threshold

Generic/Preferred Multi-Source Drug

$4.15

Not Applicable

Other

$10.35

Not Applicable

Retiree Drug Subsidy Amounts

Cost Threshold $505 $545

Cost Limit

$10,350

$11,200

(1) For a beneficiary who is not considered an “applicable beneficiary,” as defined at section

1860D-14A(g)(1) of the Act, and is not eligible for the Medicare Coverage Gap Discount

Program, this is the amount of total drug spending required to reach the OOP threshold in the

defined standard benefit. There is a 1 percent adjustment for the estimated total covered Part D

spending at catastrophic for non-applicable beneficiaries because beneficiaries take a longer time

to reach the catastrophic phase threshold when they pay less cost sharing for insulins and vaccines

(no more than a $35 copay per month’s supply of each covered insulin product and a $0 copay on

ACIP-recommended adult vaccines) under the 2024 defined standard benefit. This adjustment has

been updated from the 7 percent adjustment in the CY 2024 Advance Notice to reflect that the

low-income cost sharing subsidy, when applicable, will continue to count toward TrOOP.

(2) For a beneficiary who is an “applicable beneficiary,” as defined at section 1860D-

14A(g)(1) of the Act, and is eligible for the Medicare Coverage Gap Discount Program, this is the

estimated average amount of total drug spending required to reach the OOP threshold in the

defined standard benefit. There is a 3 percent adjustment for the estimated total covered Part D

spending at catastrophic for applicable beneficiaries. This adjustment has been updated from the 9

percent adjustment in the CY 2024 Advance Notice to account for the fact that beneficiaries take

a longer time to reach the catastrophic phase threshold when they pay less cost sharing for

insulins and vaccines (no more than a $35 copay per month’s supply of each covered insulin

136

product and a $0 copay on ACIP-recommended adult vaccines) under the 2024 defined standard

benefit.

(3) The LIS eligibility categories and corresponding cost-sharing benefits are sometimes

referred to using category codes as follows:

• Category Code 1 – Non-institutionalized FBDE beneficiaries with incomes between 100%

and 150% of FPL (beginning in CY 2024) and full-subsidy-non-FBDE beneficiaries. Note

that LIS beneficiaries that would previously fall into category code 4 fall into category

code 1 beginning in CY 2024 – see note for category code 4 below.

• Category Code 2 – Non-institutionalized FBDE beneficiaries with incomes up to 100% of

the FPL.

• Category Code 3 – FBDE beneficiaries who are institutionalized or would be

institutionalized if they were not receiving home and community-based services.

• Category Code 4 – Partial subsidy beneficiaries through CY 2023. Beneficiaries with

incomes between 135 percent and 150 percent of the FPL, who meet the resource

standards under either of sections 1860D-14(a)(3)(D) or (E) of the Act, and who would

have been eligible for the partial LIS benefit absent the enactment of the IRA, will be

eligible for the full LIS benefit. These category 4 beneficiaries will now have the same

Part D benefit parameters as beneficiaries in category 1 of the LIS. Category 2 and 3 of

the LIS remain unchanged.

(4) Per section 1860D-14(a)(1)(D)(i) of the Act, full-benefit dually eligible beneficiaries who

are receiving home and community-based services qualify for zero cost sharing if the individuals

(or couple) would have been institutionalized otherwise.

(5) The resource limits for CY 2024 will be provided via the annual HPMS memo entitled

“2024 Resource and Cost-Sharing Limits for Low-Income Subsidy (LIS)” that is expected to be

released during the usual timeframe after the September 2023 CPI has been made available by the

Bureau of Labor Statistics. Additionally, these amounts include $1,500 per person for burial

expenses. Also, beneficiaries that would have been eligible for the partial LIS benefit had the IRA

not been enacted will be eligible for the full LIS benefit if they meet either of the resource

standard described at sections 1860D-14(a)(3)(D) or (E) of the Act.

(6) The allowed amounts for total covered Part D spending at the out-of-pocket threshold

reflect the slower progression through the benefit due to the new requirements for covered insulin

products and ACIP-recommended vaccines at sections 1860D-2(b)(9) and 1860D-2(b)(8) of the

Act. For non-LIS beneficiaries, these amounts reflect that the manufacturer Coverage Gap

Discount will be calculated similarly for covered insulin products, ACIP-recommended vaccines,

and other drugs. For LIS beneficiaries, these amounts reflect that the low-income cost sharing

subsidy, when applicable, will continue to count toward TrOOP.

137

Section A. Annual Percentage Increase in Consumer Price index

Annual Percentage Increase in Consumer Price Index, September (September CPI)

Section 1860D-14(a)(4) of the Act requires CMS to use the annual percentage increase in the

CPI for the 12-month period ending in September 2023 to update the maximum copayments up

to the out-of-pocket threshold for full-benefit dually eligible enrollees with incomes not

exceeding 100 percent of the Federal Poverty Level. These copayments are increased from $1.45

per generic, preferred drug that is a multi-source drug, or biosimilar, and from $4.30 for all other

drugs in 2023 and rounded to the nearest multiple of $0.05 and $0.10 respectively.

Section B. Calculation Methodology

Annual Percentage Increase in Average Expenditures for Part D Drugs per Eligible

Beneficiary (API)

For contract years 2006 and 2007, the APIs, as defined in section 1860D-2(b)(6) of the Act, were

based on the National Health Expenditure (NHE) prescription drug per capita estimates because

sufficient Part D program data was not available. Beginning with contract year 2008, the APIs

are based on Part D program data. For the CY 2024 benefit parameters, Part D program data will

be used to calculate the annual percentage trend as follows:

 2022– 2023

 2021– 2022

= $4,916.80/$4,620.25=1.064

In the formula, the average per capita cost for August 2021 – July 2022 is calculated from actual

Part D PDE data, and the average per capita cost for August 2022 – July 2023 is calculated based

on actual Part D PDE data for prescription drug claims with service dates from August 2022 –

December 2022 and projected through July 2023.

The 2024 benefit parameters reflect the 2023 annual percentage trend, as well as an update for

revision to prior year estimates for API. Based on updated NHE prescription per capita costs and

PDE data, the annual percentage increases are now calculated as summarized by Table V-3.

Table V-3. Revised Prior Years’ Annual Percentage Increases

Year

Prior Estimates of

Annual Percentage

Trend

Revised Annual

Percentage Trend

2006

7.30%

2007

5.92%

2008

4.69%

Per section 1860D-14(a)(4)(A) of the Act, the copayments are increased from the unrounded 2023 values of $4.1508 for multi-

source generic or preferred drugs, and $10.3507 for all other drugs.

138

Year

Prior Estimates of

Annual Percentage

Trend

Revised Annual

Percentage Trend

2009

3.14%

2010

2.36%

2011

2.15%

2012

2.53%

2013

-3.14%

2014

10.12%

2015

9.89%

2016

4.02%

2017

1.87%

2018

4.05%

2019

4.92%

2020

5.06%

2021

4.69%

2022

5.80%

7.37%

Accordingly, the CY 2024 benefit parameters reflects a multiplicative update of 1.50 percent for

prior year revisions. In summary, the outlined in Section A are updated by 8.01 percent for, as

summarized by Table V-4.

Table V-4. Annual Percentage Increase

Annual percentage trend for July 2023 6.42%

Prior year revisions 1.50%

Annual percentage increase for 2024 8.01%

Note: Percentages are multiplicative, not additive. Values are carried to additional decimal places

and may not agree to the rounded values presented above.

Annual Percentage Increase for Out-of-Pocket Threshold

In accordance with section 1860D-2(b)(4)(B), we calculated the change in the out-of-pocket

threshold using the 2023 threshold value of $7,400 as our starting point. To calculate the 2024

value, we applied the 2024 API described above and rounded to the nearest $50. The resulting

2024 out-of-pocket threshold value is $8,000.

Annual Percentage Increase in Consumer Price Index, September (September CPI)

To ensure that plan sponsors and CMS have sufficient time to incorporate cost-sharing

requirements into the development of the benefit, any marketing materials, and necessary

139

systems, CMS includes in its methodology to calculate the annual percentage increase in the CPI

for the 12-month period ending in September 2023, an estimate of the September 2023 CPI

based on projections from the President’s FY2024 Budget.

The September 2022 value is from the Bureau of Labor Statistics. The annual percentage trend in

the September CPI for CY 2024 is calculated as follows:

Projected September 2023 CPI

Actual September 2022 CPI

or $308.1/$296.8=1.038

(Source: President’s FY2024 Budget and Bureau of Labor Statistics, Department of

Labor)

The CY 2024 benefit parameters reflects the CY 2023 annual percentage trend in the September

CPI of 3.81 percent, as well as a 3.87 percent multiplicative correction for the revision to last

year’s estimate. The CY 2023 annual percentage trend in the CPI can be found in Table V-5

below.

Table V-5. Cumulative Annual Percentage Increase in September CPI

Annual percentage trend for September 2023

3.81%

Prior year revisions

3.87%

Annual percentage increase for 2024

7.83%

Note: Percentages are multiplicative, not additive. Values are carried to

additional decimal places and may not agree to the rounded values presented

above.

Section C. Annual Percentage Increase in Average Expenditures for Part D Drugs per

Eligible Beneficiary (API)

Section 1860D-2(b)(6) of the Act defines the API as “the annual percentage increase in average

per capita aggregate expenditures for covered Part D drugs in the United States for Part D

eligible individuals, as determined by the Secretary for the 12-month period ending in July of the

previous year using such methods as the Secretary shall specify.” The following defined standard

Part D prescription drug benefit parameters are updated using the “annual percentage increase”:

Deductible: From $505 in 2023 and rounded to the nearest multiple of $5.

Initial Coverage Limit: From $4,660 in 2023 and rounded to the nearest multiple of $10.

Out-of-Pocket Threshold: From $7,400 in 2023 and rounded to the nearest multiple of $50.

Maximum Copayments up to the Out-of-Pocket Threshold for Certain Low-Income Full

Subsidy Eligible Beneficiaries: From $4.15 per generic, preferred drug that is a multi-source

140

drug, or biosimilar and $10.35 for all other drugs in CY 2023, rounded to the nearest multiple of

$0.05.

Table V-6. Part D Benefit Parameters for Defined Standard Benefit for 2023 and 2024 for

Non-LIS Beneficiaries

2023

2024

Deductible

Phase

Cost sharing: 100% Cost sharing: 100%

Deductible: $505 Deductible: $545

Initial

Coverage

Phase

Cost sharing: 25% Cost sharing: 25%

Initial Coverage Limit: $4,660 Initial Coverage Limit: $5,030

Coverage Gap

Applicable

Drugs:

Cost sharing:

25% (1)

Non-applicable

Drugs

Cost sharing:

25%

Applicable

Drugs

Cost sharing:

25% (1)

Non-applicable

Drugs

Cost sharing:

25%

Out-of-Pocket Threshold: $7,400 Out-of-Pocket Threshold: $8,000

Catastrophic

Coverage

Cost sharing: Greater of 5% or

$4.15 (Generic/Preferred Multi-

Source Drug) / $10.35 (Other)

Cost Sharing: 0%

(1) The 25% coinsurance for applicable drugs for non-LIS beneficiaries during the coverage

gap reflects the application of the 70% Medicare Coverage Gap Discount Program discount.

Section D. Estimated Total Covered Part D Spending at Out-of-Pocket Threshold for

Applicable Beneficiaries

For CY 2024, the estimated total gross covered prescription drug costs at the out-of-pocket

threshold for applicable beneficiaries will be calculated given the following basic assumptions:

• 100 percent beneficiary cost-sharing in the deductible phase.

• 25 percent beneficiary cost-sharing in the initial coverage phase.

• 25 percent beneficiary cost-sharing for non-applicable drugs purchased in the coverage

gap phase of the benefit.

These parameters reflect additional plan coverage required for covered insulin products under section 1860D-2(b)(9) of the

Act, as added by section 11406 of the IRA, and ACIP-recommended vaccines under section 1860D-2(b)(8) of the Act, as added

by section 11401 of the IRA.

141

• 95 percent cost-sharing for the ingredient cost and sales tax for applicable drugs

purchased in the coverage gap phase of the benefit—consisting of 25 percent beneficiary

coinsurance and 70 percent Medicare Coverage Gap Discount Program discount.

• 25 percent cost-sharing for the dispensing of applicable drugs and vaccine administration

fees not associated with ACIP-recommended vaccines drugs purchased in the coverage

gap phase of the benefit.

In this estimate, it is assumed that the dispensing and vaccine administration fees account for

0.042 percent of the gross covered brand drug costs used by non-LIS beneficiaries in the

coverage gap. Therefore, a 75 percent reduction in cost-sharing for dispensing and vaccine

administration fees results in an overall reduction of 0.030 percent to 94.970 percent in cost-

sharing for applicable (brand) drugs in the coverage gap.

The CY 2024 calculation of the estimated total gross covered prescription drug costs at out-of-

pocket (OOP) threshold for applicable beneficiaries is as follows:

(ICL+ )

100% beneficiary cost sharing in the gap

weighted gap coinsurance factor

∗    

 ($5,030 + )

$6,333.75

89.782%

∗ 1.03 = $12,447.11

• ICL is the Initial Coverage Limit equal to $5,030.

• Insulin and Vaccine Adjustment=1.03.

• 100 percent beneficiary cost sharing in the gap is the estimated total drug spending in the

gap assuming 100 percent coinsurance and is equivalent to:

(OOP threshold) – (OOP costs up to the ICL) or $8,000 − $1,666.25 = $6,333.75

Weighted gap coinsurance factor is calculated as follows:

(Brand Gross Drug Cost Below Catastrophic [GDCB] % for non-LIS × gap cost

sharing for applicable drugs) + (Generic GDCB % for non-LIS × 25% gap cost

sharing for non-applicable drugs)

(92.59% × 94.970%) + (7.41% × 25.00%) = 89.782%

o Brand GDCB % for non-LIS is the percentage of gross covered drug costs below

the OOP threshold for applicable beneficiaries (i.e., non-LIS) attributable to

applicable drugs, as reported on the 2022 PDEs.

142

o Gap cost sharing for applicable drugs is the coinsurance incurred by applicable

beneficiaries (i.e., non-LIS) for applicable drugs in the coverage gap, where:

 Coinsurance for applicable drugs is calculated as follows:

• [(percentage of gross covered brand drug costs attributable to

ingredient cost and sales tax) × (cost-sharing percentage)] + [(percentage

of gross covered brand drug costs attributable to dispensing and vaccine

administration fees not associated with ACIP-recommended vaccines) ×

(cost-sharing coinsurance percentage)]

94.970% = [(99.958% × 95%) + (0.042% × 25%)]

o Generic GDCB % for non-LIS is the percentage of gross covered drug costs

below the OOP threshold for applicable beneficiaries (i.e., non-LIS) attributable

to non-applicable drugs as reported on the 2022 PDEs.

Gap cost sharing for non-applicable drugs is the coinsurance incurred by

applicable beneficiaries (i.e., non-LIS) for non-applicable drugs in the coverage

gap.

Table V-7. Updated Total Gross Covered Drug Costs at the Out-of-Pocket Threshold

for Applicable and Non-Applicable Beneficiaries in CY 2024

2023

2024

Total Gross Covered Drug Costs at Out-of-Pocket Threshold for Non-

Applicable Beneficiaries (1) $10,516.25 $11,477.39

Estimated Total Gross Covered Drug Costs for Applicable

Beneficiaries (2)

$11,206.28

$12,447.11

(1) For a beneficiary who is not considered an “applicable beneficiary,” as defined at section

1860D-14A(g)(1) of the Act, and is not eligible for the Medicare Coverage Gap Discount

Program, this is the amount of total drug spending required to reach the OOP threshold in the

defined standard benefit.

(2) For a beneficiary who is an “applicable beneficiary,” as defined at section 1860D-

14A(g)(1) of the Act, and is eligible for the Medicare Coverage Gap Discount Program, this is the

estimated average amount of total drug spending required to reach the OOP threshold in the

defined standard benefit.

These figures have been updated from those in the CY 2024 Advance Notice to reflect the fact that beneficiaries take a longer

time to reach the catastrophic phase threshold when they pay less cost sharing for insulins and vaccines (no more than $35 copay

per month’s supply of each insulin product and $0 copay on ACIP-recommended adult vaccines) under the 2024 defined standard

benefit.

143

Section E. Retiree Drug Subsidy Amounts

Per 42 CFR 423.886(b)(3), the cost threshold and cost limit for qualified retiree prescription drug

plans are updated using the API, as defined previously in this document. The updated cost

threshold is rounded to the nearest multiple of $5 and the updated cost limit is rounded to the

nearest multiple of $50. The cost threshold and cost limit are defined as $505 and $10,350,

respectively, for plans that end in CY 2023, and as $545 and $11,200 for plans that end in CY

2024.

Table V-8. Updated Retiree Drug Subsidy Amounts in 2024

2023

2024

Retiree Drug Subsidy Amounts

Cost Threshold

$505 $545

Cost Limit

$10,350

$11,200

144

Attachment VI. Updates for Part C and D Star Ratings

Part C and D Star Ratings and Future Measurement Concepts

The Part C and D Star Ratings measure the quality of and reflect the experiences of beneficiaries

in Medicare Advantage (MA) and Prescription Drug Plans (PDPs or Part D plans), assist

beneficiaries in finding the best plan for their needs, and determine eligibility for MA Quality

Bonus Payments. The Star Ratings support CMS’ efforts to make the patient the focus in all of

our programs and to create incentives to eliminate health disparities.

The methodology for the Star Ratings system for the Part C and D programs is codified at §§

422.160 - 422.166 and 423.180 - 423.186. In the Advance Notice, we provided information and

updates as required by §§ 422.164(c)(2), (d), (e)(2) and (f)(1); 422.166(f)(2); 423.184(c)(2), (d),

(e)(2), and (f)(1); and 423.186(f)(2). We appreciate the feedback we received on potential future

measures and concepts for the Star Ratings. We reviewed the comments and will consider them

as we identify future enhancements to the Star Ratings program. Some commenters requested

additional resources to track current and future Star Ratings measures and methodological

enhancements. We will consider making additional resources available to help Part C and D

sponsors keep track of future changes.

Reminders for 2024 Star Ratings

CMS finalized the application of Tukey outlier deletion for non-CAHPS measures beginning

with the 2024 Star Ratings in the CY 2021 final rule (85 FR 33832-36).

We also finalized the

addition of the Transitions of Care and Follow-up after Emergency Department Visit for Patients

with Multiple Chronic Conditions measures to be added to the 2024 Star Ratings in the CY 2022

final rule (86 FR 5921-26). Additionally, the Plan All-Cause Readmissions measure will be

returned to the 2024 Star Ratings after being delayed due to the suspended collection of CAHPS

and HEDIS data in 2020.

We provide various datasets and reports to plan sponsors throughout the year. Part C and D

sponsors should regularly review their underlying measure data that are the basis for the Star

In the CY 2021 final rule, we finalized use of Tukey outlier deletion effective for the Star Ratings issued in October 2023 and

subsequent years. (85 FR 33833-36) In the rulemakings since that time, we have not proposed to eliminate the Tukey outlier

deletion aspect of the Star Ratings methodology. As we stated in May 2022 final rule (87 FR 27766), we will implement Tukey

outlier deletion beginning with the 2024 Star Ratings to help improve stability of cut points and prevent cut points from being

influenced by outliers. However, it appears that the sentence in §§ 422.166(a)(2)(i) and 423.186(a)(2)(i) (“Effective for the Star

Ratings issued in October 2023 and subsequent years, prior to applying mean resampling with hierarchal clustering, Tukey outer

fence outliers are removed.”) was inadvertently removed from the codified regulation text. In the Medicare Program; Contract

Year 2024 Policy and Technical Changes to the Medicare Advantage Program, Medicare Prescription Drug Benefit Program,

Medicare Cost Plan Program, Medicare Parts A, B, C, and D Overpayment Provisions of the Affordable Care Act and Programs

of All-Inclusive Care for the Elderly; Health Information Technology Standards and Implementation Specifications proposed rule

which appeared in the Federal Register on December 27, 2022 (hereinafter referred to as the December 2022 proposed rule) (87

FR 79452), we proposed a technical amendment to fix this codification error from the May 2022 final rule.

See the Announcement of Calendar Year (CY) 2022 Medicare Advantage (MA) Capitation Rates and Part C and Part D

Payment Policies, page 97.

145

Ratings and immediately alert CMS if errors or anomalies are identified so any issues can be

resolved prior to the first plan preview period.

As described at §§ 422.164(h) and 423.184(h), CMS annually sets and announces a deadline for

MA and Part D organizations to request that CMS or the Independent Review Entity (IRE)

review its Part C appeals data or CMS review its Complaints Tracking Module (CTM) data.

CMS is announcing a deadline of June 30, 2023 for all contracts to make their requests for

review of the 2022 appeals and CTM measure data for the 2024 Star Ratings. Sponsoring

organizations can view and monitor their Part C appeals timeliness and effectuation compliance

data on the Medical Appeal Search website. Sponsoring organizations should refer to the May

10, 2019 HPMS memorandum, “Complaints Tracking Module (CTM) File Layout Change and

Updated Standard Operating Procedures,” for instructions on how to request a review of CTM

data.

Measure Updates for 2024 Star Ratings

The measures that will be used to calculate the 2024 Star Ratings are listed in Table VI-1 with

information about the measure type, weight, and measurement year.

Table VI-1: 2024 Star Ratings Measures

Part C

or D

Measure

Measure Type

Weight

Measurement

Year

Improvement

Measure

Included in

the 2024 CAI

Values

Breast Cancer

Screening

Process Measure

1/1/2022 –

12/31/2022

Yes

Colorectal Cancer

Screening

Process Measure

1/1/2022 –

12/31/2022

Yes

Annual Flu Vaccine

Process Measure

3/2023 – 6/2023

Yes

Controlling Blood

Pressure

Intermediate

Outcome

Measure

1/1/2022 –

12/31/2022

Yes

Monitoring Physical

Activity

Process Measure

7/2022 –

11/2022

Yes

Special Needs Plan

(SNP) Care

Management

Process Measure

1/1/2022 –

12/31/2022

Yes

146

Part C

or D

Measure

Measure Type

Weight

Measurement

Year

Improvement

Measure

Included in

the 2024 CAI

Values

Care for Older Adults –

Medication Review

Process Measure

1/1/2022 –

12/31/2022

Yes

Care for Older Adults–

Pain Assessment

Process Measure

1/1/2022 –

12/31/2022

Yes

Osteoporosis

Management

Women who had a

Fracture

Process Measure

1/1/2022 –

12/31/2022

Yes

Diabetes Care – Eye

Exam

Process Measure

1/1/2022 –

12/31/2022

Yes

Diabetes Care – Blood

Sugar Controlled

Intermediate

Outcome

Measure

1/1/2022 –

12/31/2022

Yes

Reducing the Risk of

Falling

Process Measure

7/2022 –

11/2022

Yes

Improving Bladder

Control

Process Measure

7/2022 –

11/2022

Yes

Medication

Reconciliation Post

Discharge

Process Measure

1/1/2022 –

12/31/2022

Yes

Plan All-cause

Readmissions

Outcome

Measure

1/1/2022 –

12/31/2022

Transitions of Care

Process Measure

1/1/2022 –

12/31/2022

Follow-up after

Emergency Room Visit

Process Measure

1/1/2022 –

12/31/2022

Getting Needed Care

Patients’

Experience and

Complaints

Measure

3/2023 – 6/2023

Yes

147

Part C

or D

Measure

Measure Type

Weight

Measurement

Year

Improvement

Measure

Included in

the 2024 CAI

Values

Getting Appointments

and Care Quickly

Patients’

Experience and

Complaints

Measure

3/2023 – 6/2023

Yes

Customer Service

Patients’

Experience and

Complaints

Measure

3/2023 – 6/2023

Yes

Rating of Health Care

Quality

Patients’

Experience and

Complaints

Measure

3/2023 – 6/2023

Yes

Rating of Health Plan

Patients’

Experience and

Complaints

Measure

3/2023 – 6/2023

Yes

Care Coordination

Patients’

Experience and

Complaints

Measure

3/2023 – 6/2023

Yes

Complaints about the

Health Plan

Patients’

Experience and

Complaints

Measure

1/1/2022 –

12/31/2022

Yes

Members Choosing to

Leave the Plan

Patients’

Experience and

Complaints

Measure

1/1/2022 –

12/31/2022

Yes

Health Plan Quality

Improvement

Measure

Plan Makes Timely

Decisions about

Appeals

Measures

Capturing Access

1/1/2022 –

12/31/2022

Yes

148

Part C

or D

Measure

Measure Type

Weight

Measurement

Year

Improvement

Measure

Included in

the 2024 CAI

Values

Reviewing Appeals

Decisions

Measures

Capturing Access

1/1/2022 –

12/31/2022

Yes

Call Center – Foreign

Language Interpreter

and TTY Availability

Measures

Capturing Access

2/2023 – 5/2023

Yes

Statin Therapy for

Patients with

Cardiovascular Disease

Process Measure

1/1/2022 –

12/31/2022

Yes

Call Center – Foreign

Language

Interpreter

and TTY Availability

Measures

Capturing Access

2/2023 – 5/2023

Yes

Complaints about the

Drug Plan

Patients’

Experience and

Complaints

Measure

1/1/2022 –

12/31/2022

Yes

Members Choosing to

Leave the Plan

Patients’

Experience and

Complaints

Measure

1/1/2022 –

12/31/2022

Yes

Drug Plan Quality

Improvement

Measure

Rating of Drug Plan

Patients’

Experience and

Complaints

Measure

3/2023 – 6/2023

Yes

Getting Needed

Prescription Drugs

Patients’

Experience and

Complaints

Measure

3/2023 – 6/2023

Yes

MPF Price Accuracy

Process Measure

1/1/2022 –

9/30/2022

Yes

149

Part C

or D

Measure

Measure Type

Weight

Measurement

Year

Improvement

Measure

Included in

the 2024 CAI

Values

Medication Adherence

for Diabetes

Medications

Intermediate

Outcome

Measure

1/1/2022 –

12/31/2022

Yes

Medication Adherence

for Hypertension (RAS

antagonists)

Intermediate

Outcome

Measure

1/1/2022 –

12/31/2022

Yes

Medication Adherence

for Cholesterol

(Statins)

Intermediate

Outcome

Measure

1/1/2022 –

12/31/2022

Yes

MTM Program

Completion Rate for

CMR

Process Measure

1/1/2022 –

12/31/2022

Yes

Statin Use in Persons

with Diabetes

Process Measure

1/1/2022 –

12/31/2022

Yes

Improvement Measures (Part C & D) for the 2024 Star Ratings. Under §§ 422.164(f) and

423.184(f), improvement measures are calculated using performance measures that meet specific

conditions. Table VI-1 includes information about which measures will be used to calculate the

improvement measures for the 2024 Star Ratings. As stated in §§ 422.164(f)(4)(i) and

423.184(f)(4)(i), CMS will only include measures in the improvement calculations at the contract

level if numeric value scores are available for both the current and prior year.

2024 Star Ratings Program and the Categorical Adjustment Index

The methodology for the Categorical Adjustment Index (CAI) is described at §§ 422.166(f)(2)

and 423.186(f)(2), as well as in the annual Medicare Part C & D Star Ratings Technical Notes

available on CMS’ Part C and D Star Ratings website. As finalized at §§ 422.166(f)(2) and

423.186(f)(2), all measures identified as candidate measures will be included in the

determination of the 2024 CAI values. The measure set for the 2024 CAI (for both Part C and D)

is identified in Table VI-1.

In keeping with our commitment to transparency, a summary of the analysis of the candidate

measure set that includes the minimum, median, and maximum values for the within-contract

variation for the low-income subsidy (LIS)/dual eligible (DE) differences are posted with the

2024 CAI values on CMS’ Part C and D Star Ratings website.

150

Commenters supported continuing the CAI. There were suggestions for adding additional

measures and identifying an alternative to the CAI. We will take these suggestions into

consideration; however, Star Ratings methodological changes must be adopted through

rulemaking. We also note that certain measures, such as the CAHPS measures, are excluded

from the CAI pursuant to §§ 422.166(f)(2)(ii)(A) and 423.186(f)(2)(ii)(A) because they are

already case-mix adjusted.

Extreme and Uncontrollable Circumstances Policy for the 2024 Star Ratings

Extreme and uncontrollable circumstances such as natural disasters can directly affect

Medicare beneficiaries and providers, as well as the Parts C and D organizations that provide

beneficiaries with important medical care and prescription drug coverage. An affected

contract is identified based on these criteria:

(1) Its service area is within an “emergency area” during an “emergency period” as

defined in section 1135(g)(1) of the Act;

(2) Its service area is within a geographic area designated in a major disaster

declaration under the Stafford Act and the Secretary exercised authority under

section 1135 of the Act based on the same triggering event(s); and

(3) A certain minimum percentage (25 percent or 60 percent) of the enrollees under

the contract must reside in a Federal Emergency Management Agency (FEMA)-

designated Individual Assistance area at the time of the extreme and uncontrollable

circumstance. (See §§ 422.166(i) and 423.186(i)).

We use the start date of the incident period to determine which year of Star Ratings could be

affected, regardless of whether the incident period extends to another calendar year (§§

422.166(i) and 423.186(i)).

Under the 25 percent rules at §§ 422.166(i)(2)–(6) and 423.186(i)(2)–(5), contracts with at

least 25 percent of their service area in a FEMA-designated Individual Assistance area in

2022 will receive the higher of their measure-level rating from the current and prior Star

Ratings years for purposes of calculating the 2024 Star Ratings (thus, for 2024 Star Ratings,

affected contracts will receive the higher of their measure-level ratings from 2023 or 2024

for the applicable measures). See also 84 FR 15770–77. The numeric scores for contracts

with 60 percent or more of their enrollees living in FEMA-designated Individual Assistance

areas at the time of the extreme and uncontrollable circumstance are excluded from: (1) the

measure-level cut point calculations for non-CAHPS measures; and (2) the performance

summary and variance thresholds for the reward factor as described at §§ 422.166(i)(9)(i)

and (i)(10)(i), and 423.186(i)(7)(i) and (i)(8)(i). Table VI-2 lists the emergency areas

affected by emergency declarations first issued in 2022, as defined in section 1135 of the

Act, and the exercise of the Secretary’s authority under section 1135 of the Act.

151

Table VI-2: List of Section 1135 Waivers Issued in Relation to the FEMA Major Disaster

Declarations

Section

1135

Waiver

Date

Issued

Waiver or Modification

of Requirements Under

Section 1135 of the

Social Security Act

FEMA

Incident

Type

Affected

State

Incident

Start

Date

5/9/2022

New Mexico Wildfires

and Straight-line Winds

Wildfires

New Mexico

4/5/2022

8/2/2022

Kentucky Severe Storms,

Flooding, Landslides, and

Mudslides

Severe

Storms,

Flooding,

Landslides,

and

Mudslides

Kentucky

7/26/2022

9/20/2022

Tropical Storm/Hurricane

Fiona

Hurricane

Puerto Rico

9/17/2022

9/26/2022

Hurricane Ian

Hurricane

Florida

9/23/2022

9/30/2022

Hurricane Ian

Hurricane

South Carolina

9/25/2022

Table VI-3 lists the states and territories with Individual Assistance designations from the

FEMA major disaster declarations.

152

Table VI-3: Individual Assistance Counties and County-Equivalents in FEMA Major

Disaster Declared States/Territories

FEMA

Declaration

State

FEMA Individual Assistance Counties or County-

Equivalents

DR-4652-

New

Mexico

Colfax, Lincoln, Mora, San Miguel, Valencia

DR-4663-

Kentucky

Breathitt, Clay, Floyd, Knott, Lee, Leslie, Letcher,

Magoffin, Martin, Owsley, Perry, Pike, Whitley

DR-4671-

Puerto

Rico

Adjuntas, Aguada, Aguadilla, Aguas Buenas, Aibonito,

Anasco, Arecibo, Arroyo, Barceloneta, Barranquitas,

Bayamon, Cabo Rojo, Caguas, Camuy, Canovanas,

Carolina, Catano, Cayey, Ceiba, Ciales, Cidra, Coamo,

Comerio, Corozal, Culebra, Dorado, Fajardo, Florida,

Guanica, Guayama, Guayanilla, Guaynabo, G

urabo,

Hatillo, Hormigueros, Humacao, Isabela, Jayuya, Juana

Diaz, Juncos, Lajas, Lares, Las Marias, Las Piedras, Loiza,

Luquillo, Manati, Maricao, Maunabo, Mayaguez, Moca,

Morovis, Naguabo, Naranjito, Orocovis, Patillas, Penuelas,

Ponce, Quebradillas, Rin

con, Rio Grande, Sabana Grande,

Salinas, San German, San Juan, San Lorenzo, San

Sebastian, Santa Isabel, Toa Alta, Toa Baja, Trujillo Alto,

Utuado, Vega Alta, Vega Baja, Vieques, Villalba, Yabucoa,

Yauco

DR-4673-

Florida

Brevard, Charlotte, Collier, DeSoto, Flagler, Glades,

Hardee, Hendry, Highlands, Hillsborough, Lake, Lee,

Manatee, Monroe, Okeechobee, Orange, Osceola, Palm

Beach, Pasco, Pinellas, Polk, Putnam, Sarasota, Seminole,

St. Johns, Volusia

DR-4677-

South

Carolina

Charleston, Georgetown, Horry

Changes to Existing Star Ratings Measures for the 2023 Measurement Year and Beyond

CMS solicits feedback on new measure concepts as well as measure updates through the annual

Advance Notice and Rate Announcement process. We also provide advance notice regarding

measures considered for implementation as future Star Ratings measures. As codified at §§

422.164(c)(2)–(4), 423.184(c)(2)–(4), 422.164(d)(2), and 423.184(d)(2), new measures and

measures with substantive specification changes must be added or updated through rulemaking,

and must remain on the display page for at least two years prior to becoming a Star Ratings

measure. In addition, CMS uses the Advance Notice and Rate Announcement process to

announce non-substantive specification changes as described at §§ 422.164(d)(1) and

423.184(d)(1) and to remove measures as described at §§ 422.164(e) and 423.184(e). We

described a number of measure concepts and changes in the Advance Notice and summarize

153

significant comments on those issues here. We encourage interested parties to provide comments

directly to measure developers during their public comment periods. For example, the National

Committee for Quality Assurance (NCQA) and the Pharmacy Quality Alliance (PQA) regularly

solicit public comments on new measures, changes to existing measures, and measure

retirements.

As part of the CMS National Quality Strategy and Medicare Value-Based Care Strategy, CMS is

committed to aligning a subset of measures across all our programs and ensuring we measure

quality across the entire care continuum in a way that promotes the best, safest, and most

equitable care for all individuals. Improving alignment of measures across federal programs and

with private payers will reduce provider burden while also improving the effectiveness and

comparability of measures across quality programs. Across our CMS quality rating and value-

based care programs, where applicable, we are considering including what CMS is calling a

“Universal Foundation”

of quality measures which is a subset of measures that are aligned

across programs. This “Universal Foundation” is a building block to which programs will add

additional aligned or program-specific measures. As a start, each program is considering which

measures included in the “Universal Foundation” are not currently in their programs and the

steps to add them over time if appropriate.

Having this “Universal Foundation” will support efforts to ensure high quality care for the more

than 150 million Americans covered by our programs and serve as an alignment standard for the

rest of the health care system. The “Universal Foundation” will 1) focus provider attention, 2)

reduce provider burden, 3) allow for consistent stratification of measures to identify disparities in

care, 4) accelerate the transition to interoperable, digital quality measures, and 5) allow for cross-

comparisons across quality and value-based care programs, to better understand what drives

quality and equity improvement and what does not. The preliminary set of measures included in

the Adult “Universal Foundation” are listed in Table VI-4 with information about whether the

measures are currently in the Star Ratings program. The set of measures will evolve over time to

meet the needs of individuals served across CMS program.

Table VI-4: Preliminary Adult Universal Foundation Measures

Meaningful

Measure 2.0

Domain

Measure

Part C and D Star Ratings

Wellness and

Prevention

Colorectal Cancer Screening (HEDIS)

Currently in Star Ratings

Breast Cancer Screening (HEDIS)

Currently in Star Ratings

https://www.nejm.org/doi/full/10.1056/NEJMp2215539?query=featured_homeye

154

Meaningful

Measure 2.0

Domain

Measure

Part C and D Star Ratings

Adult Immunization Status (HEDIS)

Solicited feedback on this measure

in the Advance Notice

Chronic

Conditions

Controlling High Blood Pressure

(HEDIS)

Currently in Star Ratings

Diabetes: Hemoglobin A1c Poor

Control (>9%) (HEDIS)

Currently in Star Ratings (reversed

score so higher scores are better)

Behavioral

Health

Screening for Depression and Follow-

Up Plan (HEDIS)

Solicited feedback on this measure

in the Advance Notice

Initiation and Engagement of Substance

Use Disorder Treatment (HEDIS)

Currently on display page

Seamless care

coordination

Plan all-cause readmissions or Hospital

all-cause readmissions (HEDIS)

Currently in Star Ratings

Person-centered

care

Consumer Assessment of Healthcare

Providers and Systems (CAHPS):

Overall Rating Measures (CAHPS)

Currently in Star Ratings

Equity

Screening for Social Drivers of Health/

Social Need Screening and Intervention

(HEDIS)

Solicited feedback in the 2023

Advance Notice/Rate

Announcement about the NCQA

measure focused on Screening and

Referral to Services for Social

Needs

There was overwhelming support from commenters regarding CMS’ goal of aligning measures

across federal and private payers, and many of the commenters applauded CMS for its leadership

and efforts to streamline quality and performance measures across its many programs. Many

commenters agreed that aligning health plans and providers on a streamlined set of quality

measures could reduce administrative burden and ensure everyone is working toward the same

patient goals and outcomes. A small number of commenters suggested a variety of different

measures to add to the Universal Foundation, including measures related to women’s health

issues, tobacco use screening and intervention, patient safety, Part D medication adherence, rare

diseases, and patient activation. A commenter recommended measures that support primary care

providers, such as measures focused on relationships, access to care, comprehensiveness of care,

trust in physicians, and person-centeredness. Other commenters suggested a focus on outcome

measures rather than process measures. A handful of commenters raised issues related to the

newer measures under the Universal Foundation and whether electronic health records and the

electronic clinical data systems (ECDS) were available and ready to support accurate data

collection for these measures. Other commenters wanted more information about what programs

155

would be included in the Universal Foundation, timelines for implementing new measures in

each program included in the Universal Foundation, and how measures will be added to or

removed from the Universal Foundation over time as the focus evolves.

There was mixed reaction to some of the measures in the Universal Foundation. For Social Need

Screening and Intervention, some commenters strongly supported including this measure, but

others were concerned about data collection issues with this measure and challenges of having

clinicians screen for issues that they are not adequately able to address. We also received

feedback on the Adult Immunization Status, Initiation and Engagement of Substance Use

Disorder (SUD) Treatment, and Depression Screening and Follow-up measures; this feedback is

described below under the Display Measures section. We will take these comments into

consideration as we move forward. Any additional measures added to the Star Ratings would

need to go through rulemaking.

Optional Exclusions for HEDIS Measures (Part C). For selected HEDIS measures, plans may

choose whether or not they applied optional exclusions. NCQA reviewed all applicable HEDIS

measures to determine whether the optional exclusions could be required. NCQA is making

updates to the following Star Ratings and display measures for measurement year 2023 (2025

Star Ratings):

• Controlling Blood Pressure: The optional exclusions for pregnancy, end-stage renal

disease/dialysis/nephrectomy/kidney transplant, and non-acute inpatient admissions are

now required.

• Colorectal Cancer Screening: The optional exclusions for colorectal cancer and total

colectomy are now required.

• Kidney Health Evaluation for Patients with Diabetes: The optional exclusions for

polycystic ovary syndrome, gestational diabetes, and steroid-induced diabetes are now

required.

For all HEDIS measures that are part of the Star Ratings and display page, the optional exclusion

for enrollees who died during the measurement year became a required exclusion for

measurement year 2023. These updates would be non-substantive under § 422.164(d)(1)(i) since

they narrow the population covered under the measures. Most commenters supported making the

optional exclusions required. We have shared feedback received with NCQA.

Care for Older Adults (COA) – Pain Assessment (Part C). NCQA is considering retiring the

COA Pain Assessment indicator from the HEDIS measurement set for the following reasons: 1)

pain assessments should be multidimensional, and the current indicator cannot ensure this; 2) the

current indicator does not differentiate between acute and chronic pain; and 3) the measure also

does not assess follow up, and evidence suggests that pain assessment alone does not improve

quality of care. Additionally, the current measure is only reported for SNPs; however, a wider

population of MA enrollees would benefit from a pain assessment and follow-up measure.

Therefore, NCQA is conceptualizing a new Chronic Pain Assessment and Follow-up measure

156

described below in the section on Potential New Measure Concepts and Methodological

Enhancements for Future Years. NCQA obtained feedback from their Committee for

Performance Measurement (CPM) in September 2022 and solicited public comment on the

proposed retirement of this measure as part of the HEDIS public comment period in February

2023. Pending results of public comment, NCQA would seek approval from the CPM in May

2023 for retirement of this measure in measurement year 2025. As a reminder, in the December

2022 proposed rule, CMS proposed adding a rule at §§ 422.164(e)(1)(iii) and 423.184(e)(1)(iii)

to allow removal of a Star Ratings measure, without separate rulemaking, when a measure

steward other than CMS retires a measure.

Most commenters opposed the retirement of the COA – Pain Assessment measure until a

replacement measure has been introduced and carefully evaluated given that pain is a frequent

symptom of illness and disease in the older beneficiaries. We have shared this feedback with

NCQA for their consideration.

COA – Functional Status Assessment and Medication Review (Part C). NCQA is also

exploring the development of new measures for Functional Status Assessment and Medication

Review that may eventually replace these indicators of the COA measure and be reported for a

wider population than only enrollees of SNPs. Any potential new measures are currently planned

for development for measurement year 2025 and beyond. If new measures are developed and

implemented, NCQA would propose retirement of the existing COA measures. If NCQA retires

the existing measures, CMS would consider replacing the retired measures in the Star Ratings

with the new ones consistent with the process specified in § 422.164(c).

We received mixed reaction to the development of new measures for a wider population than

only enrollees in SNPs. Some commenters did not want the current COA measures retired until

new measures are available. Other commenters wanted more information on the value of

broadening the population for these measures, as well as more information about the specifics of

these potential new measures. We have shared this feedback with NCQA for their consideration.

Diabetes Care – Eye Exam and Diabetes Care – Blood Sugar Controlled (Part C). NCQA is

reviewing these two measures for potential updates to the existing specifications and updates that

leverage standardized electronic clinical data. NCQA is re-evaluating the approach to identify

whether an enrollee has diabetes and would be included in the denominator to reflect the

evolution of claims data coding practices, pharmacy practices, and the use of electronic clinical

data. The current method identifies enrollees if they have at least two outpatient encounters with

a diagnosis of diabetes on different dates of service or at least one inpatient encounter with a

diagnosis of diabetes or a prescription for a diabetes medication. Potential updates include 1)

simplifying the current claims-based denominator approach to identify enrollees if they have at

least two encounters (in any setting except lab) on different dates of service with a diagnosis of

diabetes; and 2) revising the current pharmacy-based denominator approach to require a diabetes

diagnosis for those enrollees identified through a dispensed diabetes medication alone (this

157

would obviate the need for the existing exclusions of polycystic ovary syndrome, gestational

diabetes, or steroid-induced diabetes). These potential clarifications for measurement year 2024

would be non-substantive under § 422.164(d)(iv) by adding clarifications for the documentation

requirements to identify enrollees with diabetes. As such, if NCQA proceeds, CMS will apply

the update to the measures beginning with the 2024 measurement year (2026 Star Ratings).

These changes would also apply to the Kidney Health Evaluation for Patients with Diabetes (Part

C) measure currently on the display page and being proposed for the 2026 Star Ratings in the

Medicare Program; Contract Year 2024 Policy and Technical Changes to the Medicare

Advantage Program, Medicare Prescription Drug Benefit Program, Medicare Cost Plan Program,

Medicare Parts A, B, C, and D Overpayment Provisions of the Affordable Care Act (ACA) and

Programs of All-Inclusive Care for the Elderly; Health Information Technology Standards and

Implementation Specifications proposed rule which appeared in the Federal Register on

December 27, 2022 (87 FR 79452, 2024 Part C and D proposed rule).

In the Advance Notice, we stated that NCQA is also evaluating the potential removal of the

hybrid reporting method for the Diabetes Care - Eye Exam and Diabetes Care - Blood Sugar

Controlled measures for measurement year 2024 and beyond. The measures would then be

specified for the Administrative or ECDS reporting methods. Currently, NCQA is not

considering removal of the hybrid method for Diabetes Care - Blood Sugar Controlled. They are

continuing to consider for the Diabetes Care–Eye Exam measure to remove hybrid reporting and

transition to administrative-only reporting. The administrative reporting method utilizes claims

data and supplemental data. The hybrid reporting method utilizes claims data, supplemental data,

and sampling to include medical record review. The removal of hybrid reporting from the

Diabetes Care–Eye Exam measure will remove the sampling component and transition the

measure to administrative-only reporting in measurement year 2024. Removing hybrid reporting

will not change the data sources that health plans can use; the change is to the reporting method

only. Health plans will no longer be able to assess performance based on a sample of members

when the hybrid method is removed, but they can perform year-round chart review and have it

audited as non-standard supplemental data to report for the measure.

NCQA is also considering excluding enrollees with bilateral eye enucleation from the Diabetes

Care – Eye Exam measure starting with measurement year 2024. This change would be non-

substantive as described at § 422.164(d)(1)(i) since it narrows the population covered by the

measure. Another potential update that NCQA is considering is the incorporation of a Glucose

Management Indicator (GMI) as an additional method to assess numerator compliance in the

existing Diabetes Care – Blood Sugar Controlled measure for measurement year 2024 or beyond.

GMI is a calculation derived from continuous glucose monitoring devices that assesses average

blood sugar values and can provide information directly to patients and physicians at more

frequent intervals. Based on guidelines from the American Diabetes Association, GMI can serve

158

as an alternative method for HbA1c for use in clinical management.

NCQA is evaluating the

inclusion of GMI alongside HbA1c as two methods to assess the numerator. In cases where

enrollees have results available for both methods in the measurement year, the most recent result

should be used (regardless of whether it is a GMI or HbA1c result). In cases where enrollees

have results available for both methods on the same day, the HbA1c result should be used. If

NCQA decides to add additional tests that meet the numerator requirements, it would be a non-

substantive update as described at § 422.164(d)(1)(iv)(A). If this update is made, NCQA is

considering renaming the measure Glycemic Status Assessment for Patients With Diabetes.

Commenters generally supported revising the specifications for identifying enrollees with

diabetes. Most of the commenters were opposed to the removal of the hybrid reporting method

for the Diabetes Care – Eye Exam measure. Some commenters noted that GMI was problematic

as it is not as precise as HbA1c. Other commenters noted that GMI is self-administered and not

all patients can afford continuous glucose monitoring devices. GMI would be an alternative to

HbA1c so would not be required. We have shared this feedback with NCQA for their

consideration as they continue to explore updates to these measures.

Breast Cancer Screening (Part C). For measurement year 2024, NCQA is considering revising

the eligible population for this measure to be more inclusive of individuals who should be

screened for breast cancer. The revised eligible population would include members 52-74 years

of age who are recommended for routine breast cancer screening, including transgender and

gender-diverse members. For example, this would include transgender men with sex assigned at

birth as female, and transgender women with sex assigned at birth as male but who have

undergone estrogen hormone therapy. The intent of this change is to ensure that all members in

need of breast cancer screening are included in the eligible population, meaningfully improving

quality of care for a population that currently experiences disparities as it relates to preventive

screenings. However, given the relatively small size of the additional population to include

transgender and non-binary individuals, this change will not meaningfully impact either the

numerator or denominator of the measure. If NCQA decides to expand the population included

in the denominator, it would be a non-substantive update as described at § 422.164(d)(1)(ii)

given less than 0.3% of adults 50 years old and older identify as transgender or nonbinary.

Most commenters supported revising the eligible population for this measure. We have shared

the feedback received with NCQA as they continue to explore updates to this measure.

Statin Use in Persons with Diabetes (SUPD) (Part D). CMS will make the following non-

substantive updates to the SUPD measure beginning with the 2024 measurement year and 2026

Star Ratings: 1) to use continuous enrollment (CE) to fully align with the PQA specifications and

to no longer adjust for member-years (MYs), and 2) to align with the PQA age criteria

American Diabetes Association Professional Practice Committee. 6. Glycemic targets: Standards of Medical Care in

Diabetes—2022. Diabetes Care 2022; 45 (Suppl. 1); S83–S96.

https://www.pewresearch.org/fact-tank/2022/06/07/about-5-of-young-adults-in-the-u-s-say-their-gender-is-different-from-

their-sex-assigned-at-birth/.

159

specifications. CMS previously solicited feedback on using the CE specifications instead of MYs

in the 2023 Advance Notice. These two changes (using CE and eligibility for the SUPD measure

by the PQA’s age criteria at the start of the measurement year) are non-substantive updates under

§ 423.184(d)(1) because they are updates with no change to the intent of the measure or the

target population.

The SUPD measure analyzes the percent of Part D beneficiaries, ages 40 to 75 years, who were

dispensed at least two diabetes medication fills that received a statin medication fill during the

measurement period. As a reminder, a higher rate for the SUPD measure indicates better

performance. CMS adapted the SUPD measure from the PQA specifications, and CMS currently

adjusts Part D enrollment based on MYs to account for beneficiaries who are enrolled for only

part of the contract year. For example, if a beneficiary is enrolled for 6 out of 12 months of the

year, they will count as only 0.5 MYs in the rate calculation. However, the current PQA

specifications use CE instead of MYs. As stated in the 2022 PQA measure manual, the

beneficiary’s index prescription start date (IPSD) begins on the earliest date of service for a

diabetes medication during the measurement year. Beneficiaries are continuously enrolled during

the measurement year with one allowable gap in enrollment which may be up to 31 days during

the measurement year.

Beginning with measurement year 2024, CMS will use CE to fully align with the PQA

specifications and to no longer adjust for MYs; this update would be non-substantive. This

update is consistent with the update to use CE for the Part D medication adherence measures as

described immediately below. In applying CE, CMS will also align with the PQA age criteria

specifications for the SUPD measure; a beneficiary will be eligible for the measure based on

their age at the start of the measurement year regardless of whether the beneficiary ages in or out

during the measurement year. This will be a non-substantive change from CMS’ current

specifications with the MY adjustment in which a beneficiary is eligible for inclusion in the

SUPD measure from the month the beneficiary meets the minimum age restriction and ending

with the month before they exceed the maximum age restriction.

We analyzed year of service (YOS) 2021 data as of January 2022 limited to contracts with a

denominator greater than 30 members and compared rates using MYs and CE. We found that 88

percent of beneficiaries included in the measure using MYs were also in the denominator using

CE. Overall, the mean SUPD rates using CE were slightly higher than rates using MYs by

around 0.3 percentage points. For MA-PDs (non-MMPs), the mean rates increased from 85.08

percent to 85.51 percent with CE. About 45 percent of MA-PD contracts’ rates stayed relatively

the same, 43 percent increased, and 12 percent decreased. For PDPs, the mean rates increased

from 82.00 percent to 82.38 percent. We found 55 percent of PDP contracts’ rates stayed

relatively the same, 43 percent increased, and 2 percent decreased. Finally, we found that there

was a slight increase in SUPD rates across MA-PDs and PDPs for both female and male

beneficiaries, as well as individuals with LIS, dual eligibility, disability, and by race (Black,

Hispanic, American Indian/Alaska Native, and Multiracial).

160

Commenters in response to the CY 2024 Advance Notice were supportive of these non-

substantive measure updates to the SUPD measure, and we will apply starting with the 2024

measurement year for the 2026 Star Ratings.

Medication Adherence for Diabetes Medication/Medication Adherence for Hypertension

(RAS Antagonists)/Medication Adherence for Cholesterol (Statins) (Part D). In the 2023

Rate Announcement, CMS solicited initial feedback on implementing risk adjustment of the

medication adherence measures based on sociodemographic (SDS) characteristics (age, gender,

dual eligibility/ LIS status, and disability status) according to the PQA specifications and

endorsed by the National Quality Forum (NQF).

Implementing SDS risk adjustment is a substantive change according to § 423.184(d)(2). CMS

proposed this change for the 2026 measurement year and 2028 Star Ratings in the 2024 Part C

and D proposed rule published on December 27, 2022 (87 FR 79616-79617). CMS also included

data analysis and additional information in the 2024 Part C and D proposed rule (87 FR 79617-

79618) to provide a more complete picture of the potential updates to the measures. The

proposed substantive change will be addressed through the rulemaking process.

We signaled that there may be a few non-substantive changes made to the adherence measures in

the Announcement of Calendar Year (CY) 2023 Medicare Advantage (MA) Capitation Rates and

Part C and Part D Payment Policies.

In addition, CMS will make the following non-substantive

changes to the three adherence measures to fully align with the current PQA measure

specifications which are endorsed by the NQF: 1) no longer adjust for MYs; instead apply the

PQA’s measure specifications to use CE as defined by the treatment period and exclude

beneficiaries with more than 1-day gap in enrollment during the treatment period and 2) no

longer adjust for inpatient (IP) or SNF stays as the PQA specifications do not include these

adjustments. As a reminder, in accordance with § 423.184(d)(1), non-substantive changes may

be adopted during or in advance of the measurement period through the Advance Notice/Rate

Announcement process. We plan to implement CE starting with the 2024 measurement year for

the 2026 Star Ratings. We plan to remove the IP/SNF stay adjustment from the adherence

measures starting with the 2026 measurement year for the 2028 Star Ratings, which is the same

time we proposed to implement the SDS risk adjustment change, but is not dependent on

finalizing the SDS risk adjustment proposal.

Commenters were supportive of the non-substantive update to the medication adherence

measures to apply CE instead of MYs to align with the PQA. CMS will implement the CE to the

medication adherence measures starting with the 2024 measurement year (2026 Star Ratings).

We received a comment requesting clarification on whether a beneficiary who has a gap in

enrollment would be excluded from the entire measurement period when applying CE. CMS will

no longer account for beneficiaries who are enrolled for only part of the measurement year in the

Please refer to the Announcement of Calendar Year (CY) 2023 Medicare Advantage (MA) Capitation Rates and Part C and

Part D Payment Policies at the following website: https://www.cms.gov/files/document/2023-announcement.pdf.

161

contract. Based on the PQA measure specifications, the treatment period for the adherence

measures begins on the index prescription start date (IPSD) and extends through whichever

comes first: the last of day of enrollment during the measurement year, death, or the end of the

measurement year. For example, a beneficiary who is CE in one contract for 6 months and then

disenrolls from that contract in July of the same measurement year would be included in the

measure calculation for that contract if they meet all of the other measure criteria, even though

the beneficiary was CE for only 6 months of the measurement year. Beneficiaries are CE during

the measurement year with an allowable one 1-day gap in enrollment during the treatment

period.

Several commenters were not supportive of the removal of the IP/SNF stay adjustment from the

medication adherence measures. Some commenters expressed concern that removal of the

IP/SNF stay adjustment would disproportionately impact plans with a higher proportion of

beneficiary stays or institutional special needs plans (I-SNPs). As a reminder, the IP/SNF stay

adjustment is not included in the PQA measure specifications. We performed contract-level

analysis of the SDS risk adjustment with and without IP/SNF stay adjustment as discussed in the

2024 Part C and D proposed rule. The majority of contracts did not have a change in their

medication adherence rates. These findings were also consistent for contracts with SNPs. We

found that more contracts with SNPs (including I-SNPs) increased in their rates or stayed the

same than decreased. CMS will move forward with this measure specification update in order to

align with the measure steward, and remove the IP/SNF stay adjustment starting with the 2026

measurement year (2028 Star Ratings).

A few commenters suggested additional medication adherence measure specification changes,

such as excluding beneficiaries in I-SNPs or residing in long-term care (LTC) facilities who are

receiving palliative care or excluding certain GLP-1 agonists which may be used for weight loss.

As a reminder, Part D does not cover medications for weight loss based on section 1927(d)(2) of

the Social Security Act. PQA is the measure steward for the adherence measures, and CMS has

shared specification comments received with the PQA.

MTM Program Completion Rate MTM Program Completion Rate for Comprehensive

Medication Review (CMR) (Part D). The data for this measure are reported by contracts to

CMS in the Health Plan Management System (HPMS) per the Part D Reporting Requirements

(OMB control number 0938-0992). Independent validation of these data is performed in

accordance with § 423.514(j) (OMB control number 0938-1115), and the results are due in

HPMS by June 30 of the year following the reporting period. Beneficiaries who are in hospice at

any point during the reporting period are excluded from this measure. The Medicare Enrollment

Database (EDB) is used to exclude beneficiaries in hospice. Starting with the 2023 reporting

period for the 2025 Star Ratings, CMS will use the EDB data to identify beneficiaries in hospice

in June after the reporting period, which aligns with when the Part D Reporting Requirements

data are pulled from HPMS. The data validation results are pulled in July of the year following

the reporting period. Commenters supported this change. This is a non-substantive change as

162

described at § 423.184(d)(1) since this change does not meaningfully impact the numerator or

denominator of the measure. In addition, the prevalence of beneficiaries in hospice is low (less

than 4% of MTM program enrollees).

Display Measures

Display measures on CMS.gov are published separately from the Star Ratings and include

measures that are transitioned from inclusion in the Star Ratings, new or updated measures

before inclusion into the Star Ratings, and informational-only measures. Organizations and

sponsors have the opportunity to preview the data for their display measures prior to release on

CMS.gov. We anticipate all 2023 display measures will continue to be shown on CMS.gov in

2024 unless noted below.

Depression Screening and Follow-Up (Part C). We solicited feedback regarding whether to

add the HEDIS Depression Screening and Follow-up for Adolescents and Adults measure to the

2026 Star Ratings display page (using data from the 2024 measurement year). As CMS’

Behavioral Health Strategy includes using quality measures to “drive health systems, providers,

practices and clinicians, and community-based providers toward delivery of high value care”

(CMS Behavioral Health Strategy | CMS), we are considering behavioral health measures that

could potentially be added to the Star Ratings program in the future through rulemaking. The

HEDIS measure, “Depression Screening and Follow-up,” measures the percentage of members

who were screened for clinical depression using a standardized instrument and, if screened

positive, received follow-up care. This aligns with the U.S. Preventive Services Task Force

recommendations regarding screening and follow-up for depression (Depression in Adults:

Screening - Healthy People 2030 | health.gov) and supports CMS’ efforts to implement the

Universal Foundation set of measures across quality programs.

Most commenters supported adding the Depression Screening and Follow-up measure to the

display page and eventually the Star Ratings. However, some commenters raised concerns about

the availability of data needed for this measure and recommended allowing time for clinical data

systems to capture this type of information. A few commenters also raised concerns about the

impact of state laws and regulations on the ability to share mental health information with

primary care providers without patient consent and the impact this may have on providing

follow-up care. CMS will take this feedback into consideration and we have also shared this

feedback with NCQA for their consideration. We plan to add this measure to the 2026 Star

Ratings display page (using data from the 2024 measurement year). For this measure to be added

to the Star Ratings, it would need to be adopted through rulemaking.

Initiation and Engagement of Substance Use Disorder (SUD) Treatment (Part C). Prior to

measurement year 2022, this measure was called Initiation and Engagement for Alcohol and

Other Drug Abuse or Dependence Treatment. This HEDIS measure is currently on our display

page. For measurement year 2022, NCQA updated the measure to change it from “member-

163

based” to “episode-based”; lengthened the negative substance use disorder (SUD) history period

from 60 days to 194 days to limit the number of members receiving ongoing treatment who

inadvertently fall into the denominator; removed emergency department visits and medically

managed withdrawal services from the negative SUD history period; removed the requirement

that a psychosocial treatment encounter accompany pharmacotherapy; and split the adult age

stratification between 18-64 years and 65+ years to better highlight any gaps in care between

different age groups. Since many individuals with SUD attempt treatment multiple times before

they are able to successfully engage, the revision of the measure to an “episode-based”

framework allows for each recovery attempt to count independently, which should result in a

more valid representation of engagement with SUD treatment for health plan populations.

Additionally, emergency department visits and withdrawal services alone are not suggestive of

ongoing or planned treatment for individuals with SUD and thus do not signal that a member is

already engaged in comprehensive care so these were removed from the measure’s negative

SUD history period. The requirement that psychosocial treatment accompany pharmacotherapy

was also removed to align with the most current clinical practice guidelines (e.g., allowing for

patients who may not accept concomitant psychosocial treatment).

We are considering potentially adding this measure to the Star Ratings in the future pending

rulemaking. This would support CMS’ efforts to implement the Universal Foundation set of

measures across quality programs. Commenters did not support adding this measure to the Star

Ratings and raised concerns about confidentiality, state and federal regulations regarding

disclosure of alcohol and SUD information without written authorization, and individuals

refusing to admit an SUD problem. We have shared this feedback with NCQA for their

consideration. We will take the comments into consideration as we consider adding this measure

to the Star Ratings. For this measure to be added to the Star Ratings, it would need to be adopted

through rulemaking.

Timely Follow-up After Acute Exacerbations of Chronic Conditions (Part C). This clinical

quality measure assesses the percentage of acute events requiring an emergency department visit

or hospitalization for one of six chronic conditions, where outpatient, non-emergent follow-up is

received within a guideline-recommended timeframe after discharge to the community for each

chronic condition:

1. Hypertension: Within 7 days

2. Asthma: Within 14 days

3. Congestive Heart Failure (CHF): Within 14 days

4. Coronary Artery Disease (CAD): Within 14 days

5. Chronic Obstructive Pulmonary Disease (COPD): Within 30 days

6. Diabetes: Within 30 days.

Follow-up care is a critical aspect of care coordination, ensuring patients understand and are

adhering to their medication regimen, providers are monitoring patients for adverse events, and

164

providers are educating patients to recognize warning signs. This measure was originally

developed using MA encounter data submitted by MA contracts reflecting care received 2014-

2016. The measure is constructed at the contract level. Details regarding measure specifications

and validation are available from NQF (NQF 34455) and the measure steward, IMPAQ

International.

The measure calculation was subsequently replicated using encounter data from 2016 to 2020.

Contract-level measure performance rates were found to be roughly stable over time with mean

and median performance rates varying from 69-70% and 71-73%, respectively, between 2016

and 2019, before declining in 2020.

CMS solicited comments on adding this measure to the display page starting with the 2024 Star

Ratings and potential future inclusion of this measure in the Part C Star Ratings pending

rulemaking. We received mixed support for this measure. While a few supported the measure

due to the importance of follow-up care, others suggested the measure is duplicative of existing

Star Ratings measures. We appreciate all comments received and will not proceed with adding

the measure to the display page at this time.

Adult Immunization Status (Part C and D). We appreciate the feedback we received from last

year’s Advance Notice on replacing the current CAHPS influenza vaccination measure with the

HEDIS influenza indicator from the Adult Immunization Status measure. Some commenters

suggested that it would be more reliable than self-reported CAHPS data, while other commenters

noted that the electronic data sources would have incomplete vaccination status data since

patients can receive vaccines in community settings with or without an insurance claim. Many

commenters cited discrepancies between HEDIS immunization data with self-reported CAHPS

data. Some commenters suggested supplementing electronic data sources with other data sources

to have more complete information. CMS will continue to take this feedback into consideration.

Any changes to the current influenza measure in Star Ratings would need to be proposed through

rulemaking.

CMS plans to add NCQA’s Adult Immunization Status measure to the 2026 display page starting

with data from the 2024 measurement year. This measure assesses the receipt of influenza,

Td/Tdap, zoster, and pneumococcal vaccines. This measure is specified for the HEDIS ECDS

Reporting Standard and captures receipt of vaccinations using data from a variety of electronic

sources such as administrative claims, immunization registries, and EHRs, among others. For

HEDIS measurement year 2023, NCQA has made a series of updates to the measure, including

updating the pneumococcal indicator to assess adults 66 and older who received any of the

following vaccines between age 19 and the end of the measurement period: pneumococcal

conjugate vaccine (PCV) 20, PCV15, PCV13, or pneumococcal polysaccharide vaccine (PPSV)

23; removing the exclusions for chemotherapy, bone marrow transplant, and

See https://www.qualityforum.org/QPS/3455.

165

immunocompromising conditions; and expanding the age range for influenza and Td/Tdap

vaccination status to Medicare enrollees age 19 and older and zoster for age 50 and older. We

also continue to consider this measure as a potential future Star Ratings measure pending

rulemaking. This measure is also part of the Universal Foundation set of measures that CMS is

considering proposing across quality programs. Commenters expressed mixed support for this

measure. Some commenters supported it and highlighted the benefits of it being a comprehensive

measure of vaccinations. Other commenters were opposed to it, expressing concerns about

obtaining complete vaccination data, inaccurate or incomplete information in immunization

registries, and enrollee resistance to vaccines given the increasingly politicized nature of

vaccinations. We have shared this feedback with NCQA. We will take these comments into

consideration as we consider adding this measure to the Star Ratings in the future through

rulemaking.

Concurrent Use of Opioids and Benzodiazepines (COB), Polypharmacy Use of Multiple

Anticholinergic Medications in Older Adults (Poly-ACH), and Polypharmacy Use of

Multiple Central Nervous System Active Medications in Older Adults (Poly-CNS) (Part D).

We announced in the 2020 Rate Announcement that these measures would be on the display

page for 2021 and 2022, and that CMS would consider adding them to the Star Ratings in the

future. In the 2024 Part C and D proposed rule (87 FR 79619-79620), CMS proposed to move

the COB, Poly-ACH, and Poly-CNS measures from the display page to the 2026 Star Ratings

(2024 measurement year). See the proposed rule for further information, and these proposals will

be addressed through the rulemaking process.

Additionally, CMS will make a non-substantive update for the 2024 measurement year to align

with the PQA measure specifications to use CE and no longer adjust for MYs. We did not

receive much feedback regarding the non-substantive update to use CE instead of MYs. One

commenter requested clarification on whether a member who has a gap in enrollment would be

excluded from the entire measurement period when applying CE. According to the PQA measure

specifications for CE for the COB and the two Polypharmacy measures, there is one allowable

gap in enrollment of up to 31 days during the measurement year. Therefore, when the enrollment

is verified monthly, the beneficiary may not have more than a one-month gap in coverage for

CE. We will implement this non-substantive update to align with the PQA measure

specifications for the 2024 measurement year.

We received measure specification comments to exclude LTC residents and add risk adjustment

to these measures. We have shared these specification-related comments with the measure

steward, PQA.

Antipsychotic Use in Persons with Dementia, Overall (APD)/Antipsychotic Use in Persons

with Dementia, in Long-Term Nursing Home Residents (APD-LTNH) (Part D). These

measures currently reported on the display page are adapted from the APD measure developed

by the PQA. The PQA recently made the following measure specification updates in their 2023

166

measure manual to the APD measure: 1) slight modification in the APD measure description; 2)

updated definition to reflect appropriate indication for antipsychotic use; 3) added major

depression diagnosis as an exclusion to the numerator; and 4) removed the “>60 cumulative days

supply” language from the denominator. In the 2023 measure manual, PQA slightly modified the

APD description from “the percentage of individuals at least 65 years of age with dementia who

received an antipsychotic medication without evidence of a psychotic disorder” to “the

percentage of individuals at least 65 years of age with dementia who received an antipsychotic

medication without evidence of an appropriate indication for an antipsychotic use.” Additionally,

the PQA updated the definition to reflect appropriate indication for antipsychotic use to align

with FDA approved uses as individuals having one or more claims with schizophrenia, bipolar

disorder, Huntington’s disease, or Tourette’s syndrome in the primary diagnosis or any other

diagnosis fields during the measurement year. Furthermore, beneficiaries taking an antipsychotic

with an FDA approved indication for treatment of major depression diagnosis (i.e., depression

resistant to treatment) is a new exclusion added to the numerator. However, the PQA developed

the following process to exclude beneficiaries with major depression diagnosis. The update to

remove beneficiaries with major depression diagnosis was approved by the PQA’s Quality

Measure Expert Panel:

1) one or more prescription claim for an antipsychotic indicated for major depression

during the measurement year. The antipsychotic medications would be based on the

PQA’s NDC lists; and

2) one or more prescription claim for an antidepressant during the measurement year. The

antidepressant medications would be based on the PQA’s NDC lists; and

3) major depression at any time during the measurement year based on the diagnosis

codes provided by the PQA.

Currently, we identify beneficiaries for the denominator who have either a dementia diagnosis

and/or two or more prescription claims with unique dates of service (DOS) and a total days’

supply greater than 60 cumulative days for a cholinesterase inhibitor or N-methyl-D-aspartate

(NMDA) receptor antagonist during the measurement year. However, in PQA’s 2023 measure

manual, PQA updates the APD measure specifications by removing the requirement for “greater

than 60 cumulative days’ supply” in the denominator since there is no known rationale for

including this requirement in addition to the 2 or more prescription claims on different dates of

service since beneficiaries can be eligible for the denominator with either prescription claims or

diagnosis. Furthermore, the removal of > 60 days’ cumulative supply would more accurately

align the APD measure with the other PQA measures.

We tested the updated PQA measure specifications for both APD measures using 2021 PDE data

with contracts with greater than 30 member-years. A total of 809 Part D contracts were included

in the APD measure analysis and 418 Part D contracts for the APD-LTNH measure. With the

167

added major depression exclusion, the numerator decreased for both APD measures. We found

that 6.0% of beneficiaries in the APD denominator population were diagnosed with major

depression and 10.1% of beneficiaries from the APD-LTNH denominator. For the APD measure,

the mean rate for all contracts improved from 8.57% to 7.10% with the updated measure

specifications. Similarly, for the APD-LTNH measure, the mean rate overall improved from

7.99% to 5.96%. As a reminder, a lower rate indicates better performance for both APD

measures. The tables below provide more information on the change in rates after applying the

updated specifications.

Table VI-5. APD Rate Distribution for Contracts with > 30 Denominator Member-Years

Contract

Type

Percentile Distributions

Number of

contracts

Mean Min p25 p50 p75 p90 Max

YOS 2021

PDE with

Current

Measure

Specifications

All

Contracts

809

8.57%

0.00%

6.09%

7.85%

10.09%

13.05%

30.75%

MAPDs

749

8.52%

0.00%

5.93%

7.67%

10.10%

13.62%

30.75%

MAPDs

(non-

MMP)

711

8.62%

0.00%

5.97%

7.72%

10.22%

13.76%

30.75%

PDPs

9.19%

5.24%

8.46%

9.04%

9.93%

11.53%

12.44%

YOS 2021

PDE with

Updated

Measure

Specifications

All

Contracts

809

7.10%

0.00%

4.75%

6.22%

8.13%

11.66%

34.06%

MAPDs

749

7.08%

0.00%

4.62%

6.05%

8.13%

11.95%

34.06%

MAPDs

(non-

MMP)

711

7.15%

0.00%

4.64%

6.12%

8.17%

12.00%

34.06%

PDPs

7.32%

2.33%

6.48%

7.30%

8.11%

9.15%

10.06%

168

Table VI-6: APD-LTNH Rate Distribution for Contracts with >30 Denominator Member-

Years

Contract

Type

Percentile Distributions

Number of

contracts

Mean Min p25 p50 p75 p90 Max

YOS 2021

PDE with

Current

Measure

Specifications

All

Contracts

418

7.99%

0.00%

4.95%

7.50%

10.37%

13.49%

20.67%

MAPDs

373

7.92%

0.00%

4.69%

7.41%

10.37%

13.49%

20.67%

MAPDs

(non-

MMP)

338

8.27%

0.00%

5.18%

7.87%

10.67%

14.15%

20.67%

PDPs

8.57%

2.93%

6.66%

7.58%

9.70%

13.05%

17.51%

YOS 2021

PDE with

Updated

Measure

Specifications

All

Contracts

418

5.96%

0.00%

3.67%

5.44%

7.81%

10.46%

27.18%

MAPDs

373

5.86%

0.00%

3.54%

5.30%

7.71%

10.27%

27.18%

MAPDs

(non-

MMP)

338

6.09%

0.00%

3.74%

5.53%

7.95%

10.88%

27.18%

PDPs

6.77%

3.03%

5.05%

5.99%

8.78%

10.65%

11.79%

Based on the results of the analysis, CMS plans to implement the updated measure specifications

on the display page for the 2023 measurement year.

One commenter requested that APD and APD-LTNH measures consider permitting the use of

low-dose, short-term use of antipsychotics in certain situations. Another commenter was

concerned with the impact and appropriateness of the APD and APD-LTNH measures in

medically complex individuals with dementia. We have shared these comments with the PQA.

The majority of commenters were supportive of the measure specification updates to the APD

and APD-LTNH measures to align with the PQA. We will implement the measure specification

updates to the APD and APD-LTNH measures for the 2023 measurement year.

Initial Opioid Prescribing - Long Duration (IOP-LD) (Part D). We began reporting the IOP-

LD measure in the 2023 display page (2021 measurement year). Currently, beneficiaries enrolled

in hospice, with a cancer diagnosis, with a sickle cell disease diagnosis, or receiving palliative

care during the measurement year or the 90 days prior to the measurement period are excluded

from the measure. However, CMS will align with current PQA measure specifications, and

therefore, these beneficiaries will be excluded from the measure during the measurement year or

90 days prior to the index prescription start date (IPSD), the earliest date of service for an opioid

medication during the measurement year. CMS plans to update the IOP-LD measure on the

display page for the 2023 measurement year.

169

The majority of commenters supported this measure specification update to the IOP-LD

measure. CMS will continue to exclude beneficiaries with an exclusion diagnosis during the

measurement year. However, with this measure specification update beginning in the 2023

measurement year, only beneficiaries with an exclusion diagnosis 90 days prior to the IPSD will

be excluded rather than excluding all beneficiaries who have an exclusion diagnosis 90 days

prior to the measurement year.

We also received one measure specification comment to exclude LTC residents from the IOP-

LD measure. We have shared this specification-related feedback with the PQA.

Medication Adherence for HIV/AIDs (Antiretrovirals) (ADH-ARV)/Antipsychotic Use in

Persons with Dementia, Overall (APD)/Antipsychotic Use in Persons with Dementia, in

Long-Term Nursing Home Residents (APD-LTNH)/Use of Opioids at High Dosage in

Persons without Cancer (OHD)/Use of Opioids from Multiple Providers in Persons without

Cancer (OMP)/Initial Opioid Prescribing -Long Duration (IOP-LD) (Part D). Similar to the

other Part D Patient Safety measures discussed above, CMS will align with the PQA measure

specifications to use CE and no longer adjust for MYs. Currently, we do not have an exact

timeline to update these display page and Patient Safety measures, but we will announce it in

advance to sponsors.

Commenters were supportive of this specification change to align with the PQA by updating the

measures from MYs to CE. We will provide more information when the timeline for these

measure changes is finalized.

Potential New Measure Concepts and Methodological Enhancements for Future Years

Health Equity (Part C and D). CMS continues to consider additional ways to advance health

equity in the Part C and D programs. CMS released confidential stratified reports to Part C and D

sponsors in HPMS in Spring 2022 to help contracts identify disparities in care by LIS/DE and

disability status for most Part C and D Star Ratings measures. Commenters were supportive of

providing confidential stratified reports to Part C and D sponsors and a few recommended

releasing the stratified reports publicly. CMS will consider releasing the stratified reports

publicly in the future. A variety of stratified reports are currently available through the CMS

Office of Minority Health website at https://www.cms.gov/about-cms/agency-

information/omh/research-and-data/stratified-reporting.

Chronic Pain Assessment and Follow-up (Part C). NCQA is exploring a new measure for

measurement year 2025 that would assess chronic pain and follow-up in Medicare enrollees age

65 and older. They are currently proposing two indicators for this measure. The first indicator

would assess if enrollees with chronic pain received a multidimensional pain assessment, and the

second indicator would assess if some type of follow-up was received among enrollees who

tested positive for pain on the multidimensional assessment. Most commenters supported this

170

measure concept but some requested more information about measure specifications. We shared

feedback received with NCQA for their consideration as they continue to develop this measure.

Cross-Cutting: Sexual Orientation and Gender Identity for HEDIS Measures (Part C).

NCQA is evaluating approaches to update applicable HEDIS measure specifications where

eligible populations are currently defined with gendered language to ensure inclusive and

gender-affirming approaches aligned with measure intent. Any potential changes to HEDIS

measures, such as Breast Cancer Screening, would be considered for measurement year 2024 or

beyond. These potential updates for measurement year 2024 or beyond would be non-substantive

under § 422.164(d)(1)(ii) because the changes are not expected to meaningfully impact the

numerator or denominator of the affected measures. Most commenters supported these updates

although some requested a longer timeframe before implementation. We shared feedback

received with NCQA.

Cross-Cutting: Identifying Chronic Conditions in HEDIS Measures (Part C). NCQA is

reevaluating how to identify those with chronic conditions (e.g., diabetes, bipolar disorder,

advanced illness) with the goal of updating the claims-based approach that is currently used

across HEDIS measures to identify conditions by incorporating clinical data. The potential

revised claims method would identify members with a condition if they have at least two

encounters with the diagnosis (in any setting except lab) on different dates of service. This would

be in place of the current method which has unneeded complexity by looking for at least two

visits (e.g., outpatient, observation, telephone, emergency department, non-acute inpatient

encounters) on different dates of service or at least one inpatient encounter or discharge with a

diagnosis. These potential updates would simplify the way conditions are identified and would

impact the following Star Ratings and display measures: Diabetes Care – Eye Exam, Diabetes

Care - Blood Sugar Controlled, Follow-up After Emergency Department Visit for Patients with

Multiple Chronic Conditions, and Kidney Health Evaluation for Patients with Diabetes. Potential

updates would also apply to how advanced illness diagnoses are identified as part of the cross-

cutting advanced illness and frailty exclusion. This exclusion is implemented in the following

Star Ratings measures: Breast Cancer Screening, Colorectal Cancer Screening, Controlling High

Blood Pressure, Diabetes Care - Eye Exam, Diabetes Care – Blood Sugar Controlled, Kidney

Health Evaluation for Patients with Diabetes, Osteoporosis Management in Women Who Had a

Fracture, Statin Therapy for Patients With Cardiovascular Disease. These potential updates for

measurement year 2024 would be non-substantive under § 422.164(d)(1)(iv) by adding

clarifications for the documentation requirements to identify enrollees with chronic conditions.

Most commenters supported these updates. We shared feedback received with NCQA for their

consideration as they make measure updates.

Blood Pressure Control Measures (Part C). NCQA is exploring the development of new blood

pressure control measures that utilize the capabilities of digital quality measures and leverage

standardized electronic clinical data. The current Controlling Blood Pressure measure included in

Part C Star Ratings assesses the percentage of members 18-85 years of age with hypertension

171

whose blood pressure was adequately controlled (<140/90 mmHg). The numerator currently

assesses if control was reached by using only the most recent blood pressure reading available.

NCQA is planning to test a new approach which takes an average of blood pressure readings

over time and will also explore alternative evidence-based blood pressure control thresholds

(<130/80 mmHg). Their testing efforts will also inform the development of an accompanying

HEDIS blood pressure control measure for patients with diabetes. The new measures are being

explored for measurement year 2025 and beyond, and if implemented, would eventually replace

the current HEDIS measures related to blood pressure. If new measures are introduced, NCQA

would propose retirement of the existing blood pressure measures.

Most commenters supported NCQA’s work to develop a new blood pressure control measure.

Some commenters requested additional clarification about the number of readings that would be

counted in the measure, whether blood pressure outliers would be removed, whether blood

pressure readings from remote patient monitoring would be included in the average, and whether

patients with only one screening in a year would be excluded from the measure. Other

commenters raised questions about alternative blood pressure thresholds under consideration.

We have shared this feedback with NCQA for their consideration as they continue to explore the

development of a new blood pressure control measure.

Kidney Health (Part C). NCQA is exploring potential measure concepts for kidney health

management related to person-centered outcomes, shared decision making, and preparedness for

kidney failure for the future. Commenters expressed strong support for these measure concepts.

We have shared this feedback with NCQA for their consideration as they continue to explore

measures related to kidney health management.

Social Connection Screening and Intervention (Part C). NCQA is continuing to work on

developing a potential new measure that assesses the percentage of members age 65 and older

who were screened using pre-specified instruments at least once during the measurement period

for social isolation, loneliness, or inadequate social support and received a corresponding

intervention if they screened positive. NCQA has begun measure development work in this area

focused on members 65 and older, because much of the evidence for screening and interventions

is focused on this age group; however, NCQA plans to explore including those under 65 in the

future. The proposed measure will have two indicators, one for social connection screening and

one for social connection intervention. This measure would be reported using electronic clinical

data, including data from electronic health records, registries, case management systems, and

administrative claims. NCQA is considering stratifying the potential measure by age (65-74, 75-

84, and 85+) and race/ethnicity. Most commenters supported these updates although some

wanted clarification about the measure and possible interventions as well as more time to

implement any changes. We have shared feedback received with NCQA.

Broadening the Mental Health Conditions Assessed by Health Outcomes Survey (HOS)

(Part C). CMS continues to explore ways to enhance the HOS to provide MA contracts with

172

useful and actionable feedback about their enrollee populations. For example, we are exploring

whether to broaden the mental health items in the survey to ensure we have the data to assess

whether enrollees with social risk factors such as low-income status are experiencing more issues

with poor mental health. This effort supports CMS’ focus on health equity.

The existing HOS mental health measures focus broadly on emotional problems with an

emphasis on depression. While depression is a significant health problem in the Medicare

population that has been linked to poor health outcomes, many older adults live with mental

health conditions beyond depression. For example, 1.2% to 15% of community samples of

persons over 60 years of age show anxiety symptoms.

We are exploring ways to measure a

broader array of mental health conditions and provide more actionable feedback and data to

health plans for quality improvement. The 2-item measure of Generalized Anxiety Disorder

(GAD-2) is used clinically as a way to screen anxiety disorders generally and Generalized

Anxiety Disorder more specifically.

In combination with the Patient Health Questionnaire-2

(PHQ-2) that is currently on the HOS and is used as a screening tool for depression, the

combined questions (GAD-2 and PHQ-2) make up the PHQ-4, which functions well as a general

mental health screening tool. Adding the GAD-2, therefore, could widen the scope of

measurement of HOS to anxiety disorders and enhance the survey’s ability to screen for mental

health needs.

CMS received mixed feedback about broadening the mental health conditions assessed by HOS.

Although over half of commenters were generally supportive, some commenters made a variety

of different suggestions such as rewording the GAD-2, considering whether there are similar

questions on the HOS, and evaluating whether the HOS is the best vehicle to assess anxiety

disorders. The remaining commenters did not support adding the GAD-2 measure to HOS, citing

reasons such as low prevalence of anxiety in older adults, limitations of HOS such as small

sample sizes, and issues with patient-reported outcome measures.

We will take this feedback into consideration as we continue to explore enhancements to HOS.

There is growing concern that anxiety disorders, similar to depression, are underrecognized and

undertreated in older adults.

Recent data from the Medicare Current Beneficiary Survey

(MCBS) and other sources suggest that the prevalence of anxiety disorders and depression

spiked during the COVID-19 pandemic. For example, one analysis of MCBS data found 24% of

adults ages 65 and older with Medicare reported anxiety or depression in August 2020, a rate that

is substantially higher than the 11% of older adults with Medicare who reported depression or

Bryant, C., Jackson, H., & Ames, D. (2008). The prevalence of anxiety in older adults: methodological issues and a review of

the literature. Journal of Affective Disorders, 109(3): 233-250.

Kroenke, K.; Spitzer, R.L.; Williams, J.B.; Monahan, P.O.; Löwe B. (2007). Anxiety disorders in primary care: prevalence,

impairment, comorbidity, and detection. Ann Intern Med, 146:317-325.

Seniors With Anxiety Frequently Don’t Get Help. Here’s Why. | Kaiser Health News (khn.org).

173

anxiety in 2018.

Rates are even higher among younger beneficiaries. An analysis of United

States Census Bureau COVID-19 Household Pulse Survey data collected between April 23, 2020

and March 1, 2021 found 43% of Medicare beneficiaries with disabilities (i.e., under age 65) had

symptoms of generalized anxiety disorder and 37% had symptoms of major depressive

disorder.

Given the impact of anxiety disorders on health, we will continue to consider how

best to measure.

Measuring Access to Mental Health Care on HOS (Part C). Since 2006, the HOS has used

seven items from the Veterans RAND 12-Item Health Survey (VR-12) to calculate mental health

summary scores. In addition, one HOS question from the VR-12 assesses change in emotional

health compared with one year ago but is not used in the calculation of the summary scores. Two

additional mental health items measure mild, moderate, or severe depression. One item assesses

memory problems.

These mental health-related questions do not address access to care, but access to mental health

care could be useful to measure for quality improvement. There are existing surveys that include

questions to assess need for and access to mental health services.

58, 59

These questions include

whether an appointment was made (or attempted) during the last 6 or 12 months, how difficult it

was to make appointments, whether psychiatric medications were prescribed or other treatment

conducted as soon as needed, or whether there were challenges in filling psychiatric

prescriptions. These types of measures could help plans assess variation in access to mental

health care.

Commenters expressed support for measuring mental health care access. However, most

commenters suggested that HOS may not be the right vehicle for this type of measure. They

believe HOS is already too long, that it has declining response rates and small sample sizes, and

that plans cannot act on the findings or track outcomes. A couple of commenters questioned

whether a self-reported measure is appropriate for tracking access to mental health care. CMS

will take this feedback into consideration as we continue to explore how best to capture

challenges Medicare beneficiaries face in accessing mental health care.

Addressing Unmet Health-Related Social Needs on HOS (Part C). In the 2023 Advance

Notice and Rate Announcement we described a new HEDIS measure focused on screening and

referral to services for social needs that NCQA refers to as the Social Need Screening and

Koma, W; True, S.; Biniek, J.F.; Cubanski, J.; Orgera, K.; Garfield, R. One in four older adults report anxiety or depression

amid the COVID-19 Pandemic. KFF, 2020, Oct 9. Accessed March 9, 2023: https://www.kff.org/medicare/issue-brief/one-in-

four-older-adults-report-anxiety-or-depression-amid-the-covid-19-pandemic/

Friedman C. (2022). The mental health of Medicare beneficiaries with disabilities during the COVID-19 pandemic. Rehabil

Psychol, 67(1):20-27.

Agency for Healthcare Research and Quality (AHRQ). (2021, May). Supplemental items for CAHPS Clinician & Group Adult

Survey 3.0/3.1: Access to Mental Health Services. https://www.ahrq.gov/cahps/surveys-guidance/item-sets/cg/suppl-

mentalhealth-cg30-adult.html.

Kyanko KA, Curry LA, Keene DE, et al. Does primary care fill the gap in access to specialty mental health care? A mixed

methods study. J Gen Intern Med. 2022:1-7.

174

Intervention (SNS-E) measure. This measure focuses on whether members were screened at least

once during the measurement year. Commenters to the Advance Notice generally supported its

use, but some requested CMS eventually go beyond this measure to include not just screening

and referrals but also access to appropriate services.

CMS is working on developing an additional measure that would complement the SNS-E

measure as we expand our work related to health equity. This new measure would be a survey-

based assessment of enrollee health-related social needs, specifically housing instability, food

insecurity, and transportation availability. Each question set would begin with an initial

screening item. The subsequent items would assess whether the respondent has received

assistance and whether a need currently exists. While the SNS-E measure aims to capture

screening and assessment by the plan and its providers, we are considering potential HOS

questions that would focus on enrollees’ perceptions of unmet needs and of the plans’ assessment

and intervention. The HOS questions would also provide additional information about ongoing

unmet needs even if the plan intervened. The HOS measure will provide important patient-

reported data that will complement the HEDIS SNS-E measure. Unlike the SNS-E, the HOS

measure will ask respondents whether they received assistance from their plan or provider and

whether they are currently struggling with unmet needs.

While most commenters see the value of addressing unmet health-related social needs and its

role in health equity, CMS received mixed feedback about adding questions regarding screening

and assistance with unmet social needs to HOS. Although more than half of commenters

supported adding related questions to HOS, other commenters raised the following concerns:

plans and providers do not have full control over which services are sought and provided;

geographic differences in the availability of community resources and interventions; adding

these items to HOS could duplicate the data from NCQA’s Social Need Screening and

Intervention (SNS-E) measure; and HOS may not be the appropriate vehicle for collecting this

information. CMS will take this feedback into consideration as we continue to explore how best

we can capture ongoing unmet needs to ensure beneficiaries are getting the support they need.

CAHPS (Part C and D).

Web Mode of Data Collection

As noted in the 2023 Advance Notice and Rate Announcement, in an effort to increase response

rates for the MA and PDP CAHPS surveys, CMS tested the effects on response rates and survey

scores of a web-based mode, as an addition to the current mixed mode protocol. The testing also

allowed for assessment of the impact of the web mode on the current MA and PDP CAHPS

survey instruments with the Agency for Healthcare Research and Quality’s (AHRQ) 5.1 Health

Plan Survey wording clarifications for explicit references to care received via telehealth (phone

or video). Commenters to the 2023 Advance Notice overwhelmingly supported the addition of a

175

web mode for the MA and PDP CAHPS survey as part of the mixed mode data collection

protocol.

In the CAHPS field test we found that for enrollees with email addresses, the web-mail-phone

protocol increased MA response rates by 4 percentage points; we found little change to response

rates for PDPs. We believe that the availability of better email addresses across all contracts will

help improve response rates overall and may help contribute to cost savings for plans in the long

run, as web responses should be less costly.

Nearly all commenters supported the addition of a web-based mode but a few were concerned

the web option would introduce bias. Based on testing and evaluation, CMS believes that there is

little to no risk of bias based on the addition of a web-based mode. The use of a three-phase

sequential multimode approach, web followed by mail followed by telephone, allows MA

enrollees choices about how to respond. It maintains or increases response rates for all groups of

Medicare enrollees and is available to those with or without broadband or telephone access.

While the increases in response rates vary slightly by enrollee characteristics, this does not create

bias, as scores from those randomized for the web-mail-phone protocol were similar to those

randomized for the mail-phone protocol in our field test. Of 39 items compared between the

web-mail-phone and mail-phone protocols, none differed in case-mix adjusted mean score at

p<0.01 and only two differed at p<0.05, a pattern consistent with chance. Thus, there is no

evidence of a mode effect on scores from the web-mail-phone protocol relative to the mail-phone

protocol.

In our testing we saw different rates of email availability by plan and, while this may influence

response rate gains, it would not bias plan scores because response by web results in scores

similar to those obtained under the mail-phone protocol. Similarly, no effect on scores over time

is anticipated. To increase the likelihood of responses, health and drug plans should maintain

accurate contact information, including email addresses when available, for their enrollees. In the

field test a majority of respondents in the web-mail-phone protocol still chose to respond by mail

or phone. Among respondents with an available email address, 79% chose to respond by mail or

phone. Further, the composition of respondents is similar in the web-mail-phone and mail-phone

protocols. We compared respondents to the web-mail-phone and mail-phone protocols by age,

sex, LIS/DE status, race/ethnicity, education, and health status, and respondents were quite

similar; the overall pattern of differences was consistent with chance.

A few commenters felt that the addition of web mode should be considered a substantive change,

suggesting it would change the denominator of the CAHPS measures. CMS disagrees that this is

a substantive change because the denominator remains enrollees with at least 6 months of

continuous enrollment at the time of sampling. With this change, the same people with Medicare

are included. The addition of web to the mail-phone survey protocol does not change the

specification for the numerator or denominator. Further, the focus of the survey questions that

176

make up the CAHPS survey has not changed, so Part C and D sponsors would be implementing

the same efforts to improve patient experiences of care.

A few commenters asked for additional information about how the web mode will be

implemented. CMS tested and will add the web mode to the existing method of administration by

using a pre-notification letter, an email with reminder (beginning in 2024) or letter survey

invitation with personalized URL to complete the survey online, up to two mailings of the

questionnaire for non-respondents, and telephone follow-up of non-respondents. This is a multi-

pronged, comprehensive survey administration protocol that avoids the weaknesses of reliance

upon mail or telephone administration alone. Additional details about the protocols can be found

in the MA and PDP CAHPS survey OMB package at https://www.cms.gov/Regulations-and-

Guidance/Legislation/PaperworkReductionActof1995/PRA-Listing-Items/CMS-R-246. Similar

to mail and phone administration, the web survey will be conducted in accordance with CMS

protocols and technical specifications. The MA and PDP CAHPS survey can only be

administered by CMS-approved vendors who have met CMS requirements for systems security

and procedures to safeguard data in a manner compliant with the Health Insurance Portability

and Accountability Act (HIPAA) as described at 45 CFR Part 160 and Part 164.

CMS also sought comment on the AHRQ’s 5.1 clarifications that explicitly added references to

in-person, phone, or video appointments to a few of the CAHPS survey items asking about

health care experiences. The survey instructions already ask the respondent to think about the

times they got health care in person, by phone, or by video call when completing the survey, so

the modified question wording just reminds the respondent of the instructions. In the field test we

did not find evidence that the 5.1 changes affect scores on the CAHPS Star Ratings measures.

Commenters generally supported the 5.1 wording changes that are inclusive of telehealth.

As discussed in the Advance Notice, we are planning to implement the web-based mode (as an

addition to the current mixed mode protocol) as well as the 5.1 wording clarifications (to

explicitly include telehealth or use terms appropriate to both telehealth and in-person visits) in

the 2024 CAHPS survey implementation used for the 2025 Star Ratings. These changes were

included in an OMB Paperwork Reduction Act package referenced above for the MA and PDP

CAHPS surveys. We note that while the 2024 Part C and D proposed rule did propose to amend

§§ 422.164(d)(1) and 423.184(d)(1) described at 87 FR 79622 to add collection of survey data

through another mode of survey administration to the non-exhaustive list of non-substantive

measure updates that can be made without rulemaking, that proposal is only a clarification. The

current regulations permit non-substantive changes like those described in this section to be done

through the Advance Notice/Rate Announcement process. As we stated in the 2024 Part C and D

proposed rule, the expansion of how data are collected is non-substantive because there is no

change to the information that is being collected; the only change is the way in which it is

collected. The CAHPS 5.1 wording changes are also non-substantive as specified at §§

422.164(d)(1)(iv)(C) and 423.184(d)(1)(iv)(C) because they reiterate the existing instructions

when answering the questions.

177

Updates to Survey Questions

Also, as noted in the 2023 Advance Notice and Rate Announcement, we tested some additional

questions for potential implementation as part of the MA and PDP CAHPS survey. The new

survey items capture more detail or test new approaches to topics covered in the current MA and

PDP CAHPS surveys (e.g., patient-provider communication, getting test results, communication

between providers), and also new topics (e.g., perceived unfair or insensitive treatment).

Commenters to the 2023 Advance Notice supported adding questions on unfair or insensitive

treatment to the survey, as long as consideration is given to survey length.

The question on unfair treatment asked whether in the last 6 months anyone from a clinic,

emergency room, or doctor’s office treated the enrollee in an unfair or insensitive way because of

their disability, age, culture or religion, language or accent, race or ethnicity, sex (female or

male), sexual orientation, gender or gender identity, or income. While few enrollees reported

experiencing unfair treatment overall, unfair treatment by health condition was most common,

followed by unfair treatment by disability and age. Across MA contracts in the field test, 9.4

percent of respondents endorsed one or more reasons for being treated in an unfair or insensitive

way.

Nearly all commenters on the CY 2024 Advance Notice supported the intent of the question on

unfair treatment, although some shared concerns about the question being added to the MA

CAHPS Survey; for example, some commenters recommended that CMS control for factors that

might affect responses. CMS is considering using the same case-mix adjustors for unfair

treatment as for other CAHPS measures. We will consider this item as a display page measure

for 2025 Star Ratings. If it were to be considered as a future Star Ratings measure, once CMS

has more experience with the measure, we would put the potential measure through the Measures

Under Consideration process, with future rulemaking used to adopt the measure for the Star

Ratings program.

We also tested modifications to the Getting Appointments and Care Quickly measure. For

example, we tested a question that would replace the current question “In the last 6 months, how

often did you see the person you came to see within 15 minutes of your appointment time?” in

the existing three-item composite measure. The replacement question that did not focus on the

exact amount of time waiting did not test well. As an alternative, we considered removing the

question related to waiting more than 15 minutes, since telehealth and type of provider may

influence how enrollees respond to this item. This would reduce the Getting Appointments and

Care Quickly measure to the existing two items:

• In the last 6 months, when you needed care right away, how often did you get care as

soon as you needed?

• In the last 6 months, how often did you get an appointment for a check-up or routine care

as soon as you needed?

178

Although this change would reduce the reliability of the measure somewhat, a two-item Getting

Appointments and Care Quickly measure would still have high reliability with a mean reliability

of 0.75.

We solicited stakeholder feedback on removing this question from the Getting

Appointments and Care Quickly measure starting with the 2024 survey administered for the

2025 Star Ratings. This change would be considered non-substantive as described at §

422.164(d)(1) since it would not change the population covered by this measure, the two existing

questions that would continue to be included in the measure, and the intent of the measure that

focuses on the individual’s experience of getting care as soon as needed. CMS received

overwhelming support for removing the 15-minute wait time question from the Getting

Appointments and Care Quickly measure. We will remove this question from Getting

Appointments and Care Quickly for the 2025 Star Ratings.

We also tested some potential alternative questions for the current questions included in the Care

Coordination measure focused on how often doctors, nurses, or health care providers explain the

results of tests, how often the explanations were easy to understand, and how often the

information was as much as was needed. We are conducting ongoing analysis of these questions

to see whether they would fit into an updated Care Coordination measure. CMS will take this

feedback into consideration as we continue to explore alternative questions.

See https://www.rand.org/pubs/technical_reports/TR653.html for a description of reliability and what is considered sufficient

reliability to discern differences among groups.

179

Attachment VII. Economic Information for the CY 2024 Rate Announcement

Below, we provide the economic information for significant provisions in the Rate

Announcement. Provisions not specifically addressed below are intended to represent a

continuation of the policies established for CY 2023 and, as a result, do not have an impact

associated with them.

Section A. Changes in the Payment Methodology for Medicare Advantage and PACE for

CY 2024

A1. Medicare Advantage and PACE non-ESRD Ratebook

The FFS growth percentage for the 2024 MA non-ESRD rates is estimated to be 2.45 percent,

and the MA growth percentage for the 2024 MA non-ESRD rates is estimated to be 1.60 percent.

The MA non-ESRD ratebook impact summarized here is calculated by comparing 2024 Part C

expenditures reflecting these growth rate assumptions to the expected 2024 Part C expenditures

assuming the MA non-ESRD ratebook remains unchanged from that finalized for 2023. The net

impact on the Medicare Trust Funds for CY 2024 is expected to be $8.1 billion. This figure

accounts for the impact of the benchmark rate cap, MA rebate, and MA EGWP policies, as well

as the portion of the difference between benchmarks and bids that the government retains and the

portion of the program costs covered by Part B premiums.

The MA growth percentage, used to calculate the 2024 PACE non-ESRD rates as well as in

development of the applicable amount used in setting MA non-ESRD rates, is estimated to

be 1.60 percent. The PACE non-ESRD ratebook impact is calculated by comparing the 2024

PACE expenditures reflecting this growth rate assumption to the expected 2024 PACE

expenditures assuming that the PACE non-ESRD ratebook remains unchanged from the CY

2023 PACE non-ESRD ratebook. The net impact on the Medicare Trust Funds for CY 2024

for the PACE ratebook change is expected to be $30 million. This figure accounts for the

portion of the program costs covered by Part B premiums.

The net impact on the Medicare Trust Funds for CY 2024 of implementing the zero-claims

adjustment in Puerto Rico is expected to be $260 million.

A2. Indirect Medical Education (IME) Phase Out

Section 161 of the Medicare Improvements for Patients and Providers Act of 2008 (MIPPA)

(Pub. L. 110-275) amended section 1853(k)(4) of the Act to require CMS to phase out indirect

medical education (IME) amounts from pre-ACA MA capitation rates, which are used to set

the cap on MA benchmarks and are used as the basis for PACE non-ESRD capitation rates.

Note that section 1894(d)(3) of the Act provides that the IME payment phase-out does not

apply to PACE capitation rates. Section 1853(n)(2)(A)(i) and (n)(2)(F) of the Act provides

that the IME phase-out is applied in developing the post-ACA MA benchmarks. Per statute,

180

the maximum incremental IME phase-out is 0.60 percent of the FFS rate per year. We

estimated the impact of the IME phase-out change between 2023 and 2024. Since the

maximum IME reduction is 8.4 percent in 2023 and 9.0 percent in 2024, we calculate the

impact as the difference for those counties with IME percentages of at least 8.4 percent, with

the maximum impact of 0.6 percent (i.e., the difference between 8.4 and 9.0 percent). Also,

since the IME reduction to MA benchmarks is increasing, the impact is considered to be a net

savings to the Medicare Trust Funds.

In payment year 2024, there are no counties that have IME amounts greater than 8.4 percent of

the FFS rate. Since all counties have IME amounts less than 8.4 percent of their respective FFS

rates, there is no impact by the change in the IME phase-out percentage in 2024. For the ESRD

ratebook, all IME amounts used for MA ESRD rates are less than 8.4 percent of the FFS rate, so

there is no impact from the IME phase-out change on the ESRD ratebook for 2024.

Note that the statutorily prescribed methodology for calculating the IME phase-out in 2024 is the

same as that provided by statute for CY 2023; we are providing this impact assessment for

informational purposes.

A3. Medicare Advantage and PACE ESRD Ratebooks

The FFS growth percentage for the 2024 MA ESRD rates is estimated to be 2.27 percent. The

impact on the MA and PACE ESRD ratebooks is calculated by comparing projected 2024 Part C

expenditures with this growth rate assumption to the expected 2024 Part C expenditures with the

assumption that the MA and PACE ESRD ratebooks would have been unchanged from those

finalized for 2023. The net impact on the Medicare Trust Funds for CY 2024 is expected to be

$440 million. This figure accounts for the portion of the program costs covered by Part B

premiums.

A4. CMS-HCC Risk Adjustment Model

For CY 2024, CMS is finalizing an updated CMS-HCC risk adjustment model for organizations

other than PACE, with a 3-year phase in beginning in CY 2024 when the risk scores will be

calculated as the sum of 33% of the risk score calculated with the updated model (the 2024

model) and 67% of the risk score calculated with the current model (the 2020 model). The CY

2024 impact on MA risk scores of the finalized CMS-HCC risk adjustment model with a 3-year

phase in starting at 33% is projected to be -2.16%, which represents a $7.6 billion net savings to

the Medicare Trust Fund in 2024.When estimating the impact of the proposed model, the impact

takes into account the portion of the difference between benchmarks and bids that the

government retains and the portion of the program costs covered by Part B premiums.

Section B. Changes in the Payment Methodology for Medicare Part D for CY 2024

181

A5. ESRD Risk Adjustment

For CY 2024, CMS is continuing the use of the ESRD risk adjustment models implemented in

CY 2023. Therefore, no economic impact is applicable.

A6. Frailty Adjustment for FIDE SNPs

For CY 2024, CMS is calculating frailty scores for FIDE SNPs by blending 67% of the frailty

scores calculated with the frailty factors used for CY 2023 (associated with the 2020 CMS-HCC

model) and 33% of the frailty scores calculated with the updated frailty factors being finalized

for CY 2024 that do not include the CAHPS survey weight (associated with the finalized 2024

CMS-HCC risk adjustment model). To calculate impacts, CMS utilized the survey results from

the 2021 HOS / HOS-M to estimate frailty scores based on the frailty factors used for CY 2023

(the current model) and the frailty factors being finalized for CY 2024 (the updated model), and

blended them as is being finalized. The CY 2024 impact of transitioning to frailty scores

calculated using the updated frailty factors, relative to CY 2023, is a change in frailty scores of -

0.58%, which represents a net savings of less than $10 million dollars to the Medicare Trust

Funds in 2024.

A7. MA Coding Pattern Adjustment

For CY 2024, we will continue to apply the statutory minimum coding pattern difference

adjustment (5.90%). There is no change in policy from CY 2023, and we applied the same factor

for CY 2023, therefore the year-over-year impact is zero.

A8. Normalization

The normalization factors serve to offset the trend in risk scores and maintain a 1.0 average FFS

risk score. For CY 2024, for the CMS-HCC risk adjustment models with a 2019 or 2020

denominator, CMS will calculate the normalization factors using a five-year linear slope

methodology and updated average FFS risk scores for 2018 through 2022, but continuing to

exclude the 2021 risk score as was done for the CY 2023 normalization factor. For the CMSHCC

risk adjustment models with a 2015 denominator and the RxHCC models, CMS will calculate

the normalization factors using a five-year linear slope methodology and historical FFS risk

scores (2016 through 2020). Since normalization is applied to risk scores to maintain the same

average risk scores in each program year-over-year, the impact of normalization is zero.

B1. Part D Risk Adjustment Model

For CY 2024, we are continuing the use of the RxHCC risk adjustment model that was

implemented in CY 2023. Therefore, no economic impact is applicable.

182

B2. Annual Percentage Increase for Part D Parameters

The methodology for updating other Part D parameters for CY 2024 remains unchanged from

that used for CY 2023. As a result, updating the other Part D parameters does not have an impact

on the Medicare Trust Fund alone; the impact of such parameter updates is dependent on the

behavior and bid assumptions of Part D plan sponsors.

183

Attachment VIII. CMS-HCC Risk Adjustment Factors & Predictive Ratio Tables

Table VIII-1. 2024 CMS-HCC Model Relative Factors for Continuing Enrollees

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

Female

0-34 Years

0.238

0.346

0.454

0.948

35-44 Years

0.288

0.332

0.420

0.810

45-54 Years

0.340

0.384

0.404

1.031

55-59 Years

0.385

0.421

0.424

0.949

60-64 Years

0.436

0.502

0.414

0.881

65-69 Years

0.330

0.435

0.365

1.188

70-74 Years

0.395

0.506

0.423

1.119

75-79 Years

0.465

0.596

0.485

0.965

80-84 Years

0.524

0.665

0.544

0.862

85-89 Years

0.624

0.775

0.618

0.750

90-94 Years

0.737

0.869

0.738

0.627

95 Years or Over

0.742

0.877

0.835

0.481

Male

0-34 Years

0.106

0.191

0.306

0.826

35-44 Years

0.154

0.204

0.261

0.719

45-54 Years

0.215

0.293

0.300

0.991

55-59 Years

0.283

0.410

0.353

0.989

60-64 Years

0.345

0.504

0.374

0.917

65-69 Years

0.332

0.531

0.375

1.275

70-74 Years

0.396

0.626

0.417

1.224

75-79 Years

0.502

0.714

0.498

1.319

80-84 Years

0.571

0.789

0.565

1.238

85-89 Years

0.664

0.907

0.615

1.135

90-94 Years

0.800

0.993

0.712

0.946

95 Years or Over

0.896

1.058

0.904

0.825

Medicaid and Originally Disabled Interactions

Originally Disabled, Female

0.228

0.160

0.103

Originally Disabled, Male

0.135

0.158

0.075

Medicaid

0.130

184

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

Disease Coefficients

HCC1

HIV/AIDS

0.301

0.213

0.397

0.237

0.196

0.109

1.322

HCC2

Septicemia, Sepsis, Systemic

Inflammatory Response

Syndrome/Shock

0.500 0.598 0.649 0.780 0.447 0.591 0.605

HCC6

Opportunistic Infections

0.381

0.763

0.588

0.833

0.518

0.685

0.728

HCC17

Cancer Metastatic to Lung,

Liver, Brain, and Other Organs;

Acute Myeloid Leukemia

Except Promyelocytic

4.209 3.995 3.896 4.235 3.946 4.103 1.952

HCC18

Cancer Metastatic to Bone,

Other and Unspecified

Metastatic Cancer; Acute

Leukemia Except Myeloid

2.341 2.486 2.277 2.537 2.166 2.403 1.110

HCC19

Myelodysplastic Syndromes,

Multiple Myeloma, and Other

Cancers

1.798 1.989 1.563 1.661 1.520 1.554 0.957

HCC20

Lung and Other Severe Cancers

1.136

0.978

1.166

1.173

1.214

1.067

0.672

HCC21

Lymphoma and Other Cancers

0.671

0.540

0.654

0.739

0.627

0.618

0.493

HCC22

Bladder, Colorectal, and Other

Cancers

0.363 0.366 0.382 0.409 0.410 0.351 0.314

HCC23

Prostate, Breast, and Other

Cancers and Tumors

0.186 0.233 0.196 0.218 0.203 0.237 0.197

HCC35

Pancreas Transplant Status

0.949

1.393

1.117

0.573

1.117

2.740

1.106

HCC36

Diabetes with Severe Acute

Complications

0.166 0.191 0.186 0.235 0.166 0.210 0.280

HCC37

Diabetes with Chronic

Complications

0.166 0.191 0.186 0.235 0.166 0.210 0.280

HCC38

Diabetes with Glycemic,

Unspecified, or No

Complications

0.166 0.191 0.186 0.235 0.166 0.210 0.280

HCC48

Morbid Obesity

0.186

0.144

0.300

0.178

0.164

0.118

0.442

HCC49

Specified Lysosomal Storage

Disorders

9.256 13.778 2.833 6.399 3.269 7.771 1.528

185

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

HCC50

Amyloidosis, Porphyria, and

Other Specified Metabolic

Disorders

0.648 0.883 0.555 0.789 0.435 0.529 0.362

HCC51

Addison's and Cushing's

Diseases, Acromegaly, and

Other Specified Endocrine

Disorders

0.510 0.606 0.634 0.654 0.313 0.393 0.620

HCC62

Liver Transplant

Status/Complications

0.376 0.184 0.261 0.409 0.571 0.271 0.593

HCC63

Chronic Liver Failure/End-

Stage Liver Disorders

0.962 1.032 1.102 1.209 0.861 1.101 0.894

HCC64

Cirrhosis of Liver

0.447

0.383

0.475

0.414

0.391

0.270

0.378

HCC65

Chronic Hepatitis

0.185

0.248

0.101

0.220

0.156

0.189

0.378

HCC68

Cholangitis and Obstruction of

Bile Duct Without Gallstones

0.388 0.383 0.085 0.354 0.391 0.270 0.090

HCC77

Intestine Transplant

Status/Complications

1.172 6.301 5.039 6.161 5.039 5.039 5.089

HCC78

Intestinal

Obstruction/Perforation

0.326 0.534 0.382 0.548 0.478 0.688 0.380

HCC79

Chronic Pancreatitis

0.357

0.574

0.525

0.799

0.444

0.709

0.218

HCC80

Crohn's Disease (Regional

Enteritis)

0.550 0.635 0.490 0.651 0.479 0.603 0.374

HCC81

Ulcerative Colitis

0.244

0.285

0.201

0.286

0.205

0.237

0.258

HCC92

Bone/Joint/Muscle/Severe Soft

Tissue Infections/Necrosis

0.479 0.529 0.611 0.632 0.471 0.539 0.556

HCC93

Rheumatoid Arthritis and Other

Specified Inflammatory

Rheumatic Disorders

0.617 0.470 0.439 0.384 0.405 0.288 0.297

HCC94

Systemic Lupus Erythematosus

and Other Specified Systemic

Connective Tissue Disorders

0.268 0.239 0.237 0.250 0.224 0.196 0.297

HCC107

Sickle Cell Anemia (Hb-SS)

and Thalassemia Beta Zero

0.457 1.449 0.610 1.939 0.303 1.569 0.692

186

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

HCC108

Sickle Cell Disorders, Except

Sickle Cell Anemia (Hb-SS)

and Thalassemia Beta Zero;

Beta Thalassemia Major

0.146 0.386 0.103 0.408 0.303 0.416 0.098

HCC109

Acquired Hemolytic, Aplastic,

and Sideroblastic Anemias

1.144 1.815 1.048 1.541 1.009 1.514 0.529

HCC111

Hemophilia, Male

4.639

30.706

15.539

31.424

11.201

32.199

6.310

HCC112

Immune Thrombocytopenia

and Specified Coagulation

Defects and Hemorrhagic

Conditions

0.450 0.640 0.460 0.634 0.574 0.708 0.516

HCC114

Common Variable and

Combined Immunodeficiencies

2.262 2.598 2.016 2.670 2.137 2.789 0.691

HCC115

Specified Immunodeficiencies

and White Blood Cell

Disorders

0.565 0.692 0.438 0.498 0.302 0.613 0.691

HCC125

Dementia, Severe

0.341

0.296

0.438

0.367

0.401

0.345

HCC126

Dementia, Moderate

0.341

0.296

0.438

0.367

0.401

0.345

HCC127

Dementia, Mild or Unspecified

0.341

0.296

0.438

0.367

0.401

0.345

HCC135

Drug Use with Psychotic

Complications

0.424 0.637 0.702 1.181 0.522 0.922 0.297

HCC136

Alcohol Use with Psychotic

Complications

0.424 0.637 0.502 1.181 0.522 0.922 0.297

HCC137

Drug Use Disorder,

Moderate/Severe, or Drug Use

with Non-Psychotic

Complications

0.424 0.365 0.502 0.471 0.394 0.348 0.297

HCC138

Drug Use Disorder, Mild,

Uncomplicated, Except

Cannabis

0.423 0.264 0.502 0.384 0.355 0.348 0.297

HCC139

Alcohol Use Disorder,

Moderate/Severe, or Alcohol

Use with Specified Non-

Psychotic Complications

0.242 0.207 0.478 0.250 0.308 0.159 -

187

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

HCC151

Schizophrenia

0.511

0.380

0.591

0.414

0.501

0.304

0.449

HCC152

Psychosis, Except

Schizophrenia

0.484 0.290 0.579 0.255 0.501 0.247 0.208

HCC153

Personality Disorders;

Anorexia/Bulimia Nervosa

0.396 0.290 0.420 0.255 0.464 0.232 0.199

HCC154

Bipolar Disorders without

Psychosis

0.351 0.166 0.349 0.126 0.314 0.108 0.199

HCC155

Major Depression, Moderate or

Severe, without Psychosis

0.299 0.166 0.316 0.126 0.269 0.108 0.199

HCC180

Quadriplegia

1.125

0.986

1.068

1.095

1.311

1.399

0.735

HCC181

Paraplegia

0.942

0.648

0.859

0.832

0.883

0.852

0.563

HCC182

Spinal Cord Disorders/Injuries

0.478

0.368

0.402

0.308

0.401

0.316

0.270

HCC190

Amyotrophic Lateral Sclerosis

and Other Motor Neuron

Disease, Spinal Muscular

Atrophy

1.175 1.792 1.427 3.642 0.640 1.243 0.628

HCC191

Quadriplegic Cerebral Palsy

0.855

0.743

0.393

0.466

0.840

0.104

HCC192

Cerebral Palsy, Except

Quadriplegic

0.314 0.129 - 0.067 0.220 0.104 -

HCC193

Chronic Inflammatory

Demyelinating Polyneuritis and

Multifocal Motor Neuropathy

1.692 1.427 0.957 0.939 1.149 0.850 0.913

HCC195

Myasthenia Gravis with

(Acute) Exacerbation

2.909 3.633 2.153 3.323 2.690 1.779 1.837

HCC196

Myasthenia Gravis without

(Acute) Exacerbation and Other

Myoneural Disorders

0.516 0.642 0.503 0.507 0.427 0.248 0.486

HCC197

Muscular Dystrophy

0.426

0.632

0.369

0.681

0.162

0.145

0.292

HCC198

Multiple Sclerosis

0.647

0.908

0.791

1.143

0.569

0.770

0.226

HCC199

Parkinson and Other

Degenerative Disease of Basal

Ganglia

0.615 0.517 0.634 0.504 0.474 0.354 0.219

188

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

HCC200

Friedreich and Other

Hereditary Ataxias; Huntington

Disease

0.279 0.208 0.165 0.281 0.050 0.428 -

HCC201

Seizure Disorders and

Convulsions

0.245 0.196 0.233 0.170 0.245 0.202 0.131

HCC202

Coma, Brain

Compression/Anoxic Damage

0.543 0.238 0.721 0.279 0.549 0.309 0.097

HCC211

Respirator

Dependence/Tracheostomy

Status/Complications

0.879 0.878 1.981 1.418 1.022 0.590 1.570

HCC212

Respiratory Arrest

0.370

0.510

0.573

0.662

0.409

0.493

0.258

HCC213

Cardio-Respiratory Failure and

Shock

0.370 0.510 0.573 0.662 0.409 0.493 0.258

HCC221

Heart Transplant

Status/Complications

1.053 0.999 1.412 1.781 0.880 1.371 0.840

HCC222

End-Stage Heart Failure

2.505

5.770

2.927

6.612

3.009

6.106

0.826

HCC223

Heart Failure with Heart Assist

Device/Artificial Heart

2.505 5.770 2.927 6.612 3.009 6.106 0.826

HCC224

Acute on Chronic Heart Failure

0.360

0.442

0.406

0.537

0.311

0.411

0.217

HCC225

Acute Heart Failure (Excludes

Acute on Chronic)

0.360 0.442 0.406 0.537 0.311 0.411 0.217

HCC226

Heart Failure, Except End-

Stage and Acute

0.360 0.442 0.406 0.537 0.311 0.411 0.217

HCC227

Cardiomyopathy/Myocarditis

0.189

0.200

0.173

0.198

0.145

0.186

0.189

HCC228

Acute Myocardial Infarction

0.252

0.254

0.493

0.517

0.324

0.407

0.310

HCC229

Unstable Angina and Other

Acute Ischemic Heart Disease

0.240 0.254 0.325 0.458 0.278 0.315 0.310

HCC238

Specified Heart Arrhythmias

0.299

0.296

0.407

0.304

0.293

0.261

0.245

HCC248

Intracranial Hemorrhage

0.239

0.180

0.377

0.332

0.313

0.183

0.081

HCC249

Ischemic or Unspecified Stroke

0.239

0.180

0.377

0.277

0.299

0.172

0.081

HCC253

Hemiplegia/Hemiparesis

0.387

0.320

0.437

0.390

0.437

0.403

HCC254

Monoplegia, Other Paralytic

Syndromes

0.321 0.172 0.292 0.365 0.290 0.335 -

189

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

HCC263

Atherosclerosis of Arteries of

the Extremities with Ulceration

or Gangrene

1.118 1.066 1.432 1.276 1.007 1.056 0.696

HCC264

Vascular Disease with

Complications

0.455 0.520 0.498 0.461 0.513 0.622 0.338

HCC267

Deep Vein Thrombosis and

Pulmonary Embolism

0.294 0.431 0.445 0.568 0.338 0.498 0.245

HCC276

Lung Transplant

Status/Complications

2.531 1.583 2.210 2.292 2.961 1.277 3.085

HCC277

Cystic Fibrosis

0.998

2.818

1.340

3.760

0.650

3.829

0.873

HCC278

Idiopathic Pulmonary Fibrosis

and Lung Involvement in

Systemic Sclerosis

0.818 1.209 0.791 1.640 0.650 0.937 0.873

HCC279

Severe Persistent Asthma

0.818

0.842

0.594

0.808

0.650

0.804

0.873

HCC280

Chronic Obstructive Pulmonary

Disease, Interstitial Lung

Disorders, and Other Chronic

Lung Disorders

0.319 0.209 0.390 0.281 0.321 0.234 0.312

HCC282

Aspiration and Specified

Bacterial Pneumonias

0.440 0.362 0.538 0.269 0.409 0.173 0.353

HCC283

Empyema, Lung Abscess

0.204

0.131

0.074

HCC298

Severe Diabetic Eye Disease,

Retinal Vein Occlusion, and

Vitreous Hemorrhage

0.336 0.364 0.323 0.319 0.327 0.301 0.545

HCC300

Exudative Macular

Degeneration

0.596 0.366 0.370 0.255 0.459 0.380 0.196

HCC326

Chronic Kidney Disease, Stage

0.815 0.927 0.985 0.946 0.965 1.050 0.958

HCC327

Chronic Kidney Disease,

Severe (Stage 4)

0.514 0.523 0.565 0.661 0.484 0.447 0.462

HCC328

Chronic Kidney Disease,

Moderate (Stage 3B)

0.127 0.179 0.116 0.181 0.140 0.178 0.145

HCC329

Chronic Kidney Disease,

Moderate (Stage 3, Except 3B)

0.127 0.179 0.116 0.181 0.140 0.178 0.145

190

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

HCC379

Pressure Ulcer of Skin with

Necrosis Through to Muscle,

Tendon, or Bone

1.965 2.140 2.580 2.570 2.349 2.349 1.420

HCC380

Chronic Ulcer of Skin, Except

Pressure, Through to Bone or

Muscle

1.078 1.091 1.422 1.285 1.268 1.378 0.839

HCC381

Pressure Ulcer of Skin with

Full Thickness Skin Loss

1.075 1.091 1.379 1.192 1.136 1.089 0.423

HCC382

Pressure Ulcer of Skin with

Partial Thickness Skin Loss

0.838 0.994 1.029 0.935 0.845 0.937 0.343

HCC383

Chronic Ulcer of Skin, Except

Pressure, Not Specified as

Through to Bone or Muscle

0.646 0.707 0.890 0.707 0.660 0.654 0.343

HCC385

Severe Skin Burn

1.291

0.234

2.362

0.857

0.204

HCC387

Pemphigus, Pemphigoid, and

Other Specified Autoimmune

Skin Disorders

0.406 0.302 0.658 0.622 0.477 0.498 0.125

HCC397

Major Head Injury with Loss of

Consciousness > 1 Hour

0.199 0.150 0.349 0.190 0.128 0.052 0.085

HCC398

Major Head Injury with Loss of

Consciousness < 1 Hour or

Unspecified

0.199 0.150 0.349 0.190 0.128 0.052 0.085

HCC399

Major Head Injury without

Loss of Consciousness

0.199 0.150 0.349 0.190 0.128 0.052 0.085

HCC401

Vertebral Fractures without

Spinal Cord Injury

0.522 0.605 0.622 0.559 0.538 0.412 0.231

HCC402

Hip Fracture/Dislocation

0.467

0.561

0.570

0.499

0.527

0.089

HCC405

Traumatic Amputations and

Complications

0.598 0.577 0.799 0.844 0.639 0.698 0.284

HCC409

Amputation Status, Lower

Limb/Amputation

Complications

0.598 0.562 0.799 0.844 0.604 0.623 0.284

191

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

HCC454

Stem Cell, Including Bone

Marrow, Transplant

Status/Complications

1.068 0.452 1.326 0.608 1.338 0.416 1.596

HCC463

Artificial Openings for Feeding

or Elimination

0.673 0.914 0.891 0.947 0.526 0.853 0.634

Disease Interactions

DIABETES_HF

Diabetes*Heart Failure

0.112

0.023

0.183

0.041

0.164

0.053

0.209

HF_CHR_

LUNG

Heart Failure*Chronic Lung

Disorder

0.078 0.062 0.109 0.097 0.140 0.108 0.145

HF_KIDNEY

Heart Failure*Kidney

0.176

0.314

0.194

0.420

0.140

0.328

CHR_LUNG_CARD

_RESP_FAIL

Chronic Lung

Disorder*Cardiorespiratory

Failure

0.254 0.242 0.340 0.275 0.329 0.270 0.331

HF_HCC238

Heart Failure*Specified Heart

Arrhythmias

0.077 0.257 0.140 0.372 0.135 0.314 -

gSubUseDisorder_

gPsych_

Substance Use

Disorder*Psychiatric

- 0.087 - 0.152 - 0.149 -

Disabled/Disease Interactions

DISABLED_

Disabled, Heart Failure - - - - - - 0.488

DISABLED_

ULCER_

Disabled, Skin Ulcer - - - - - - 0.537

DISABLED_

CANCER

Disabled, Cancer - - - - - - 0.367

DISABLED_

NEURO_

Disabled, Neurological - - - - - - 0.154

DISABLED_

CHR_LUNG

Disabled, Chronic Lung

Disorder

- - - - - - 0.278

Payment HCC Counts

1 payment HCCs

2 payment HCCs

3 payment HCCs

4 payment HCCs

5 payment HCCs

0.050

0.088

0.049

0.095

0.016

0.105

192

Variable Description Label

Community,

NonDual,

Aged

Community,

NonDual,

Disabled

Community,

FBDual,

Aged

Community,

FBDual,

Disabled

Community,

PBDual,

Aged

Community,

PBDual,

Disabled

Institutional

6 payment HCCs

0.102

0.223

0.071

0.245

0.096

0.191

7 payment HCCs

0.188

0.380

0.160

0.472

0.207

0.435

8 payment HCCs

0.316

0.440

0.267

0.607

0.345

0.581

9 payment HCCs

0.444

0.750

0.353

0.841

0.345

0.823

D10P

10 or more payment HCCs

0.728

1.431

0.746

1.471

0.901

1.268

0.373

NOTES:

1. The denominator used is $10,402.34.

2. In the “disease interactions” and “disabled interactions,” the variables are defined as follows:

Cancer = HCCs 17-23

Cardiorespiratory Failure = HCCs 211-213

Chronic Lung Disorder = HCCs 276-280

Diabetes = HCCs 35-38

Heart Failure = HCCs 221-226

Kidney = HCCs 326-329

Neurological = HCCs 180-192, 195, 196, 198, 199

Psychiatric = HCCs 151-155

Skin Ulcer = HCCs 379-382

Specified Heart Arrhythmias = HCC 238

Substance Use = HCCs 135-139

SOURCE: 2018-2019 100% Medicare data.

193

Table VIII-2. 2024 CMS-HCC Model Relative Factors for Aged and Disabled New Enrollees

Non-Medicaid &

Non-Originally

Disabled

Medicaid &

Non-Originally

Disabled

Non-Medicaid &

Originally

Disabled

Medicaid &

Originally

Disabled

Female

0-34 Years

0.711

1.025

35-44 Years

0.950

1.303

45-54 Years

1.155

1.415

55-59 Years

1.152

1.289

60-64 Years

1.212

1.396

65 Years

0.532

0.986

1.212

1.599

66 Years

0.532

0.990

1.276

1.599

67 Years

0.557

1.004

1.276

1.599

68 Years

0.584

1.004

1.276

2.021

69 Years

0.625

1.004

1.276

2.021

70-74 Years

0.694

1.043

1.276

2.021

75-79 Years

0.901

1.128

1.276

2.021

80-84 Years

0.988

1.342

1.276

2.021

85-89 Years

1.287

1.563

1.287

2.021

90-94 Years

1.287

1.712

1.287

2.021

95 Years or Over

1.287

1.712

1.287

2.021

Male

0-34 Years

0.409

0.738

35-44 Years

0.669

1.264

45-54 Years

0.906

1.420

55-59 Years

0.984

1.477

60-64 Years

1.057

1.542

65 Years

0.567

1.182

1.057

1.727

66 Years

0.576

1.234

1.155

1.959

67 Years

0.617

1.319

1.155

1.959

68 Years

0.678

1.367

1.155

1.959

69 Years

0.684

1.455

1.297

1.959

70-74 Years

0.808

1.455

1.297

1.959

75-79 Years

1.049

1.455

1.297

2.813

80-84 Years

1.245

1.503

1.297

2.813

85-89 Years

1.516

1.682

1.516

2.813

90-94 Years

1.516

1.981

1.516

2.813

95 Years or Over

1.516

1.981

1.516

2.813

NOTES:

3. The denominator used is $10,402.34.

4. For payment purposes, a new enrollee is a beneficiary who did not have 12 months of Part B eligibility in the data

collection year. CMS-HCC new enrollee models are not based on diagnoses, but include factors for different age and

sex combinations by Medicaid and the original reason for Medicare entitlement.

SOURCE: 2018-2019 100% Medicare data.

194

Table VIII-3. 2024 CMS-HCC Model Relative Factors for New Enrollees in Chronic Condition

Special Needs Plans (C-SNPs)

Non-Medicaid &

Non-Originally

Disabled

Medicaid &

Non-Originally

Disabled

Non-Medicaid &

Originally

Disabled

Medicaid &

Originally

Disabled

Female

0-34 Years

1.332

1.655

35-44 Years

1.332

1.655

45-54 Years

1.536

1.965

55-59 Years

1.536

1.989

60-64 Years

1.628

2.030

65 Years

0.900

1.285

1.708

2.085

66 Years

0.900

1.285

1.708

2.085

67 Years

0.940

1.351

1.719

2.148

68 Years

1.004

1.351

1.745

2.148

69 Years

1.020

1.467

1.755

2.233

70-74 Years

1.195

1.634

1.929

2.295

75-79 Years

1.419

1.885

2.032

2.484

80-84 Years

1.612

2.061

2.239

2.735

85-89 Years

1.833

2.250

2.239

2.735

90-94 Years

2.016

2.400

2.239

2.735

95 Years or Over

2.016

2.400

2.239

2.735

Male

0-34 Years

1.206

1.485

35-44 Years

1.206

1.485

45-54 Years

1.472

1.845

55-59 Years

1.552

1.994

60-64 Years

1.642

2.035

65 Years

0.944

1.422

1.642

2.035

66 Years

0.944

1.422

1.659

2.150

67 Years

0.985

1.477

1.659

2.195

68 Years

1.005

1.477

1.659

2.195

69 Years

1.065

1.477

1.677

2.246

70-74 Years

1.216

1.735

1.807

2.316

75-79 Years

1.496

1.952

2.039

2.487

80-84 Years

1.704

2.194

2.155

2.573

85-89 Years

1.924

2.403

2.346

2.573

90-94 Years

2.142

2.403

2.346

2.573

95 Years or Over

2.142

2.403

2.346

2.573

NOTES:

1. The denominator used is $10,402.34.

2. For payment purposes, a new enrollee is a beneficiary who did not have 12 months of Part B eligibility in the data

collection year. CMS-HCC new enrollee models are not based on diagnoses, but include factors for different age and

sex combinations by Medicaid and the original reason for Medicare entitlement.

SOURCE: 2018-2019 100% Medicare data.

195

Table VIII-4. 2024 CMS-HCC Model with Disease Hierarchies

CMS-HCC If the Disease Group is listed in this column…

…Then drop the

CMS-HCC listed in

this column

CMS-HCC Hierarchical Condition Category Label

Cancer Metastatic to Lung, Liver, Brain, and Other Organs; Acute Myeloid

Leukemia Except Promyelocytic

18, 19, 20, 21, 22, 23

Cancer Metastatic to Bone, Other and Unspecified Metastatic Cancer;

Acute Leukemia Except Myeloid

19, 20, 21, 22, 23

Myelodysplastic Syndromes, Multiple Myeloma, and Other Cancers

20, 21, 22, 23

Lung and Other Severe Cancers

21, 22, 23

Lymphoma and Other Cancers

22, 23

Bladder, Colorectal, and Other Cancers

Pancreas Transplant Status

36, 37, 38

Diabetes with Severe Acute Complications

37, 38

Diabetes with Chronic Complications

Liver Transplant Status/Complications

63, 64, 65, 68

Chronic Liver Failure/End-Stage Liver Disorders

64, 65, 68, 202

Cirrhosis of Liver

65, 68

Intestine Transplant Status/Complications

78, 80, 81

Crohn's Disease (Regional Enteritis)

Rheumatoid Arthritis and Other Specified Inflammatory Rheumatic

Disorders

107

Sickle Cell Anemia (Hb-SS) and Thalassemia Beta Zero

108

111

Hemophilia, Male

112

114

Common Variable and Combined Immunodeficiencies

115

125

Dementia, Severe

126, 127

126

Dementia, Moderate

127

135

Drug Use with Psychotic Complications

136, 137, 138, 139

136

Alcohol Use with Psychotic Complications

137, 138, 139

137

Drug Use Disorder, Moderate/Severe, or Drug Use with Non-Psychotic

Complications

138, 139

138

Drug Use Disorder, Mild, Uncomplicated, Except Cannabis

139

151

Schizophrenia

152, 153, 154, 155

152

Psychosis, Except Schizophrenia

153, 154, 155

153

Personality Disorders; Anorexia/Bulimia Nervosa

154, 155

154

Bipolar Disorders without Psychosis

155

180

Quadriplegia

181, 182, 253, 254

181

Paraplegia

182, 254

191 Quadriplegic Cerebral Palsy

180, 181, 182, 192,

253, 254

192 Cerebral Palsy, Except Quadriplegic

180, 181, 182, 253,

254

195

Myasthenia Gravis with (Acute) Exacerbation

196

211

Respirator Dependence/Tracheostomy Status/Complications

212, 213

212

Respiratory Arrest

213

196

CMS-HCC If the Disease Group is listed in this column…

…Then drop the

CMS-HCC listed in

this column

CMS-HCC Hierarchical Condition Category Label

221 Heart Transplant Status/Complications

222, 223, 224, 225,

226, 227

222 End-Stage Heart Failure

223, 224, 225, 226,

227

223

Heart Failure with Heart Assist Device/Artificial Heart

224, 225, 226, 227

224

Acute on Chronic Heart Failure

225, 226, 227

225

Acute Heart Failure (Excludes Acute on Chronic)

226, 227

226

Heart Failure, Except End-Stage and Acute

227

228

Acute Myocardial Infarction

229

248

Intracranial Hemorrhage

249

253

Hemiplegia/Hemiparesis

254

263

Atherosclerosis of Arteries of the Extremities with Ulceration or Gangrene

264, 383, 409

276

Lung Transplant Status/Complications

277, 278, 279, 280

277

Cystic Fibrosis

278, 279, 280

278

Idiopathic Pulmonary Fibrosis and Lung Involvement in Systemic Sclerosis

279, 280

279

Severe Persistent Asthma

280

282

Aspiration and Specified Bacterial Pneumonias

283

326

Chronic Kidney Disease, Stage 5

327, 328, 329

327

Chronic Kidney Disease, Severe (Stage 4)

328, 329

328

Chronic Kidney Disease, Moderate (Stage 3B)

329

379

Pressure Ulcer of Skin with Necrosis Through to Muscle, Tendon, or Bone

380, 381, 382, 383

380

Chronic Ulcer of Skin, Except Pressure, Through to Bone or Muscle

381, 382, 383

381

Pressure Ulcer of Skin with Full Thickness Skin Loss

382, 383

382

Pressure Ulcer of Skin with Partial Thickness Skin Loss

383

397

Major Head Injury with Loss of Consciousness > 1 Hour

202, 398, 399

398

Major Head Injury with Loss of Consciousness < 1 Hour or Unspecified

202, 399

405

Traumatic Amputations and Complications

409

How Payments are Made with a Disease Hierarchy

EXAMPLE: If a beneficiary triggers HCCs 195 (Myasthenia Gravis with (Acute) Exacerbation) and 196 (Myasthenia Gravis

without (Acute) Exacerbation and Other Myoneural Disorders), then HCC 196 will be dropped. In other words, payment will

always be associated with the HCC in column 1 if an HCC in column 3 also occurs during the same collection period. Therefore,

the organization’s payment will be based on HCC 195 rather than HCC 196.

197

Table VIII-5. Predictive Ratios by Deciles of Predicted Risk (sorted low to high):

Non-Dual, Aged (Age >=65) Continuing Enrollee

2014/2015 Sample 2018/2019 Sample

Deciles

2020 Model 2020 Model 2024 Model

Improvement in

Predictive Risk

Entire sample

1.000

0.968

1.000

First (lowest) decile

0.968

0.902

0.977

Second decile

0.983

0.938

0.981

Third decile

0.996

0.940

1.026

Fourth decile

0.989

0.958

1.003

Fifth decile

1.003

0.977

0.995

Sixth decile

1.002

0.970

0.993

Seventh decile

1.005

0.983

0.996

Eighth decile

1.003

0.982

0.996

Ninth decile

1.003

0.987

1.006

Tenth (highest)

1.003

0.963

1.003

Top 5%

1.000

0.942

1.000

Top 1%

0.984

0.917

0.987

Top 0.1%

0.959

0.879

0.967

Table VIII-6. Predictive Ratios by Deciles of Predicted Risk (sorted low to high):

Non-Dual, Disabled (Age <65) Continuing Enrollee

2014/2015 Sample 2018/2019 Sample

Deciles

2020 Model 2020 Model 2024 Model

Improvement in

Predictive Risk

Entire sample

1.000

0.979

1.000

First (lowest) decile

1.090

1.100

0.932

Second decile

0.959

0.975

0.990

Third decile

0.982

0.964

0.983

Fourth decile

0.982

0.977

1.011

Fifth decile

0.952

0.968

0.955

Sixth decile

0.997

0.965

0.997

Seventh decile

0.983

0.972

0.997

Eighth decile

1.008

1.004

1.002

Ninth decile

1.028

1.013

1.022

Tenth (highest)

1.001

0.959

1.004

Top 5%

0.991

0.935

0.998

Top 1%

0.999

0.922

0.981

Top 0.1%

0.979

0.874

0.960

198

Table VIII-7. Predictive Ratios by Deciles of Predicted Risk (sorted low to high):

Full Benefit Dual, Aged (Age >=65) Continuing Enrollee

2014/2015 Sample 2018/2019 Sample

Deciles

2020 Model 2020 Model 2024 Model

Improvement in

Predictive Risk

Entire sample

1.000

1.002

1.000

First (lowest) decile

0.969

0.949

0.996

Second decile

1.006

0.980

1.029

Third decile

0.988

1.012

1.015

Fourth decile

0.994

0.996

0.983

Fifth decile

1.006

1.017

0.986

Sixth decile

1.000

1.006

0.997

Seventh decile

1.004

1.012

0.992

Eighth decile

1.003

1.014

1.002

Ninth decile

1.002

1.009

1.002

Tenth (highest)

1.001

0.991

1.003

Top 5%

1.004

0.983

1.002

Top 1%

0.978

0.938

0.979

Top 0.1%

0.915

0.844

0.919

Table VIII-8. Predictive Ratios by Deciles of Predicted Risk (sorted low to high):

Full Benefit Dual, Disabled (Age <65) Continuing Enrollee

2014/2015 Sample 2018/2019 Sample

Deciles

2020 Model 2020 Model _ 2024 Model

Improvement in

Predictive Risk

Entire sample 1.000 0.988 1.000 -

First (lowest) decile 1.076 1.008 0.967

Second decile 1.016 1.004 1.053

Third decile 0.893 0.869 0.904

Fourth decile 0.940 0.957 0.970

Fifth decile 0.992 0.985 1.005

Sixth decile 0.999 1.010 1.005

Seventh decile 1.020 0.995 1.013

Eighth decile 1.019 0.999 0.996

Ninth decile 1.008 1.014 1.016

Tenth (highest) 1.002 0.983 1.002

Top 5% 0.996 0.974 0.995

Top 1% 0.984 0.954 0.983

Top 0.1% 0.873 0.986 1.007

199

Table VIII-9. Predictive Ratios by Deciles of Predicted Risk (sorted low to high):

Partial Benefit Dual, Aged (Age >=65) Continuing Enrollee

2014/2015 Sample 2018/2019 Sample

Deciles

2020 Model 2020 Model 2024 Model

Improvement in

Predictive Risk

Entire sample

1.000

0.992

1.000

First (lowest) decile

0.998

0.942

1.000

Second decile

0.998

0.987

1.023

Third decile

0.977

0.933

0.999

Fourth decile

0.987

0.992

1.001

Fifth decile

0.999

0.989

0.976

Sixth decile

1.004

1.016

0.983

Seventh decile

1.003

1.013

1.006

Eighth decile

1.006

1.017

1.000

Ninth decile

1.006

1.021

1.009

Tenth (highest)

0.999

0.968

1.000

Top 5%

0.994

0.951

1.000

Top 1%

0.999

0.931

0.985

Top 0.1%

0.981

0.870

0.981

Table VIII-10. Predictive Ratios by Deciles of Predicted Risk (sorted low to high):

Partial Benefit Dual, Disabled (Age <65) Continuing Enrollee

2014/2015 Sample 2018/2019 Sample

Deciles

2020 Model 2020 Model 2024 Model

Improvement in

Predictive Risk

Entire sample

1.000

0.988

1.000

First (lowest) decile

0.935

0.878

0.989

Second decile

1.020

1.023

0.896

Third decile

0.988

0.955

1.045

Fourth decile

0.979

0.991

1.002

Fifth decile

0.982

0.979

0.996

Sixth decile

0.999

0.988

1.003

Seventh decile

1.011

1.012

0.999

Eighth decile

1.025

1.032

0.996

Ninth decile

1.010

1.019

1.022

Tenth (highest)

0.996

0.963

1.000

Top 5%

0.989

0.944

0.997

Top 1%

1.002

0.939

0.981

Top 0.1%

1.076

0.932

0.968

200

Table VIII-11. Predictive Ratios by Deciles of Predicted Risk (sorted low to high):

Institutional Continuing Enrollee

2014/2015 Sample 2018/2019 Sample

Deciles

2020 Model 2020 Model 2024 Model

Improvement in

Predictive Risk

Entire sample

1.000

0.951

1.000

First (lowest) decile

0.858

0.788

0.824

Second decile

0.959

0.877

0.932

Third decile

0.995

0.928

0.977

Fourth decile

1.000

0.949

1.011

Fifth decile

1.022

0.968

1.029

Sixth decile

1.023

0.976

1.035

Seventh decile

1.026

0.982

1.028

Eighth decile

1.020

0.975

1.028

Ninth decile

1.015

0.970

1.014

Tenth (highest)

0.989

0.952

0.992

Top 5%

0.984

0.939

0.978

Top 1%

0.967

0.900

0.918

Top 0.1%

0.954

0.865

0.859

NOTES:

1. “Improvement in Predictive Risk” compares the distance the predictive ratios are from 1.0 for the 2024 model and

2020 model with a 2018 – 2019 sample.

2. For example, a green arrow indicates that the predictive ratio for any specific decile for the 2024 model is closer to 1.0

than the predictive ratio for the 2020 model with a 2018 – 2019 sample, and vice-versa.