HCC coding and risk adjustment statistics

In March 2026, 35.2 million of 64.2 million Medicare beneficiaries with Parts A and B were enrolled in Medicare Advantage, or 55%.

Scope: KFF uses March 2026 CMS dashboard data for the national penetration calculation. Dashboard records may change as CMS reconciles them.

Figure 1 and the opening enrollment section.

Medicare Advantage enrollment grew by about 1.1 million beneficiaries, or 3%, between March 2025 and March 2026.

Scope: This is March-to-March administrative enrollment.

Text immediately following Figure 1.

In March 2026, individual plans open for general enrollment held 21.4 million Medicare Advantage enrollees, equal to 61% of total enrollment.

Scope: The share fell from 62% in March 2025 even as enrollment rose by 0.2 million. KFF excludes plan-county cells with fewer than 11 beneficiaries, about 400,000 people.

Figure 3 and the opening paragraph of the individual-plan section.

In March 2026, nearly 8.2 million Medicare beneficiaries were enrolled in Special Needs Plans, equal to 23% of Medicare Advantage enrollment.

Scope: Special Needs Plan enrollment rose by more than 900,000 from March 2025 and represented 85% of net Medicare Advantage enrollment growth. KFF excludes plan-county cells with fewer than 11 beneficiaries.

Figure 4 and the first two paragraphs of the Special Needs Plan section.

Chronic Condition Special Needs Plan enrollment reached about 1.7 million people in March 2026, up 45%, or about 518,000 enrollees, from March 2025.

Scope: C-SNPs represented 20% of all Special Needs Plan enrollment. KFF excludes plan-county cells with fewer than 11 beneficiaries.

C-SNP paragraph below Figure 4.

Dual-Eligible Special Needs Plans held 78% of Special Needs Plan enrollment in March 2026.

Scope: The share was down from 83% in March 2025. KFF excludes plan-county cells with fewer than 11 beneficiaries.

D-SNP paragraph below Figure 4.

UnitedHealth Group and Humana together accounted for 46% of Medicare Advantage enrollment in March 2026.

Scope: UnitedHealth held 26%, or 9.3 million enrollees, and Humana held 20%, or 7 million. KFF excludes plan-county cells with fewer than 11 beneficiaries.

Figure 6 and the parent-organization market-share section.

Three quarters of individual Medicare Advantage prescription drug plan enrollees paid no supplemental premium in 2026, and the enrollment-weighted supplemental premium was $15 per month.

Scope: The $15 average includes zero-premium enrollees, excludes the standard Part B premium, and applies to individual MA-PD plans.

Figure 1 and the opening premium section.

The average 2026 Medicare Advantage out-of-pocket limit was $5,421 for in-network services, while the PPO combined in-network and out-of-network average was $9,825.

Scope: These are plan limits, not actual enrollee spending.

Figure 3 and surrounding text.

Nearly all Medicare Advantage enrollees, 99%, were in plans requiring prior authorization for at least some services in 2026.

Scope: This is the share enrolled in plans with a requirement, not a request or denial rate.

Figure 9 and the prior-authorization section.

CMS projected that its final 2026 Medicare Advantage payment policies would raise plan payments by an average of 5.06%, or more than $25 billion, compared with 2025.

Scope: Plan-specific effects vary, and the 5.06% projection excludes the underlying Medicare Advantage coding trend.

Opening paragraph and Net Payment Impact table.

CMS expected the underlying Medicare Advantage coding trend to increase average risk scores by 2.10% in 2026.

Scope: This industry-wide average is not included in the 5.06% expected average revenue change.

Net Payment Impact, footnote 4.

MedPAC expected Medicare's capitated Medicare Advantage payments to average $16,242 per beneficiary in 2026.

Scope: The expected average includes rebate payments.

Chapter summary, page 344.

MedPAC projected Medicare Advantage rebate payments of $2,660 per beneficiary in 2026, equal to 15% of total Medicare Advantage payments and more than double the 2018 amount.

Scope: Based on plan bid projections. Rebates finance benefits, premium reductions, administrative expenses, and profit.

Chapter summary, page 344.

MedPAC projected total Medicare payments to Medicare Advantage plans of $615 billion in 2026.

Scope: This Part C projection includes enrollees with ESRD and excludes separately determined Part D payments. It is not finalized spending.

Chapter summary, page 347.

MedPAC estimated that Medicare would spend 14%, or $76 billion, more for Medicare Advantage enrollees in 2026 than if those beneficiaries were in fee-for-service Medicare.

Scope: The estimate includes the ESRD population and is not an estimate of plan profit or administrative expense.

Chapter summary, pages 346 to 347.

MedPAC estimated that higher Medicare Advantage spending would add about $11 billion to Part B premium payments in 2026, roughly $175 per beneficiary for the year or $14.61 per month.

Scope: This is MedPAC's aggregate estimate across all Part B beneficiaries, including beneficiaries in fee-for-service Medicare.

Chapter summary, page 346.

CMS completed the three-year phase-in of the 2024 CMS-HCC model in 2026, with 100% of non-PACE risk scores calculated under that model.

Scope: This applies to organizations other than PACE and describes the model blend, not a RAF value.

Part C Risk Adjustment Model section.

For 2026 PACE payments, CMS blended 10% of the 2024 CMS-HCC model with 90% of the 2017 CMS-HCC model.

Scope: This blend applies only to PACE organizations.

Part C Risk Adjustment Model for PACE Organizations.

The 2024 CMS-HCC model has 266 total HCCs and 115 payment HCCs, compared with 204 total HCCs and 86 payment HCCs in the 2020 model.

Scope: These are model-category counts, not counts of ICD-10-CM codes.

Table 2-2, pages 18 to 19.

In the model-calibration code universe, the 2024 CMS-HCC model mapped 7,770 of 73,926 ICD-10-CM codes to payment HCCs, or 10.5%, compared with 9,797 codes, or 13.3%, in the 2020 model.

Scope: Uses fiscal year 2022 and 2023 codes available at calibration, not the current fiscal year 2026 code set.

Table 2-2, page 18.

CMS reported that the 2024 model newly mapped 209 ICD-10-CM codes to payment HCCs and removed 2,236 codes from payment HCC mappings.

Scope: The added and removed counts are directional mapping changes, not a net code-count calculation.

Table 2-2, page 18.

CMS noted that the total ICD-10-CM code set grew from 73,926 codes in the model-calibration data to 74,044 codes in fiscal year 2024.

Scope: These totals include codes that do not map to payment HCCs.

Table 2-2 note, page 19.

Across its six community segments, the 2024 CMS-HCC model's individual-level R-squared values ranged from 0.1159 to 0.1889, the highest range among the historical segmented CMS-HCC models CMS compared.

Scope: CMS says subgroup predictive ratios, not R-squared, are the primary accuracy measure for the model's intended use.

Section 2.1, page 14.

CMS recalibrated the 2024 CMS-HCC model with 2018 fee-for-service diagnoses and 2019 expenditures, replacing the 2014 diagnosis and 2015 expenditure years used for the prior model.

Scope: These are calibration data years, not 2026 dates of service.

Section 2.2.1, page 17.

MedPAC projected that favorable selection would increase 2026 Medicare Advantage payments by roughly 11% above estimated fee-for-service spending, accounting for $57 billion of the $76 billion total payment difference.

Scope: This is a model-based estimate before plan management or coding-intensity effects.

Chapter summary, page 347.

Before the CMS coding adjustment, MedPAC projected Medicare Advantage risk scores would be about 10% higher than scores for similar fee-for-service beneficiaries in 2026.

Scope: This is a projection, not an audit error rate. MedPAC estimated the V28 phase-in reduced coding intensity by 2.9 percentage points in each year from 2024 through 2026.

Chapter summary, page 348.

After the 5.9% minimum CMS coding adjustment, MedPAC projected a residual risk-score difference of about 4%, adding an estimated $22 billion to 2026 Medicare Advantage payments.

Scope: The $22 billion is a MedPAC estimate, not a recovery amount or a finding of improper payment.

Chapter summary, page 348.

A peer-reviewed study estimated that private Medicare plan enrollees generated diagnosis-based risk scores 6% to 16% higher than the same enrollees would have generated in fee-for-service Medicare.

Scope: This is a counterfactual estimate, not a claim that every plan or enrollee was upcoded by that amount.

Abstract.

For each $1 of potential Medicare Advantage revenue associated with coding intensity, one study found plan bids fell by $0.10 to $0.19, premiums fell by $0.11 to $0.16, and $0.21 to $0.45 went toward lower medical loss ratios.

Scope: Regression associations using 2008 to 2015 administrative data and prescription-drug utilization as an independent health-risk proxy.

Abstract, Principal Findings, and Tables 1 and 2.

A study associated chart reviews with $2.3 billion in added Medicare Advantage payments in 2015, equal to a 3.7% spending increase, with 10% of contracts accounting for 42% of the added payments.

Scope: Cross-sectional analysis of 14,021,692 beneficiaries in 510 contracts. The authors could not validate added diagnoses against medical records.

Abstract, Results, and Figure 1.

CMS announced targets to expand its medical-coder team from 40 to about 2,000, annual audits from about 60 plans to about 550 eligible plans, and records reviewed per plan from 35 to a range of 35 to 200.

Scope: These were announced targets. The source does not prove CMS achieved them.

Key Elements of the Plan, Workforce Expansion and Increased Audit Volume.

CMS's March 2026 schedule listed six planned RADV audit initiations covering payment years 2020 through 2025, with starts scheduled from March 2026 through April 2027.

Scope: CMS says dates are subject to change and payment years may be audited out of sequence.

One-page audit schedule table.

OIG estimated that diagnoses found only on health risk assessments or linked chart reviews generated $7.5 billion in 2023 Medicare Advantage risk-adjusted payments for 1.7 million enrollees.

Scope: Based on 2022 encounter data. OIG did not find that all $7.5 billion was improper.

Estimated Risk-Adjusted Payments, page 1, Exhibit 2.

In-home health risk assessments and linked chart reviews accounted for 63%, or $4.7 billion, of OIG's $7.5 billion payment estimate.

Scope: The linked chart-review amount includes reviews connected to assessments conducted in any setting.

Page 1, Exhibit 2 and note.

OIG estimated $1,869 in risk-adjusted payments per in-home health risk assessment record, compared with $365 per facility-based assessment record.

Scope: Estimated from 2022 encounter data. Record shares are not enrollee shares.

Page 2, Exhibits 3 and 4.

Thirteen health conditions accounted for $5.6 billion, or 75%, of OIG's health-risk-assessment-related payment estimate.

Scope: The lack of another service record for the diagnosis does not prove every diagnosis was invalid.

Top Health Conditions, page 3, Exhibit 5.

Twenty Medicare Advantage companies generated 80% of OIG's $7.5 billion health-risk-assessment payment estimate while covering 50% of Medicare Advantage enrollees.

Scope: A disproportionate payment share is not by itself proof of miscoding. OIG listed the 20 companies in Appendix C, Exhibit C-1.

Top MA Companies, pages 4 to 5, Exhibit 7.

OIG found 19,028 Medicare Advantage enrollees with no 2022 encounter record other than one health risk assessment, associated with an estimated $81.9 million in payments.

Scope: Encounter data can be incomplete. This finding does not establish that every enrollee received no care.

MA Enrollees With No Other 2022 Service Records, pages 7 to 8, Exhibit 11.

In a targeted Priority Health audit, OIG found unsupported diagnoses in 252 of 300 sampled enrollee-years and calculated $828,010 in sample net overpayments.

Scope: Do not convert this targeted sample result into a plan-wide error rate.

Findings, pages 7 to 8, and Appendix C, Tables 4 and 5.

OIG's 2026 acute-stroke audit examined a high-risk population of 240,401 enrollees across 554 Medicare Advantage organizations, associated with $477,460,953 in payments for the selected HCC.

Scope: The population was selected for a specific record pattern and does not represent all stroke coding.

How We Conducted This Audit, page 5, and Appendix A, page 11.

None of the 97 relevant sampled acute-stroke HCCs in OIG's targeted audit were validated: 93 were unsupported by supplied records and four records could not be located.

Scope: The sample was deliberately restricted to a high-risk pattern. Missing records differ from reviewed records that did not support a diagnosis.

Findings, pages 5 to 6, and Appendix C, Tables 3 and 4.

The U.S. Bureau of Labor Statistics estimated 194,720 employed medical records specialists in May 2025.

Scope: Wage-and-salary workers only. Self-employed workers and unpaid family workers are excluded.

Table 1, PDF page 13, Medical records specialists row.

The May 2025 national mean wage for medical records specialists was $27.30 per hour and $56,790 annually.

Scope: Annual mean wages use a 2,080-hour work year and exclude overtime, benefits, and certain bonuses.

Table 1, PDF page 13, Medical records specialists row.

The May 2025 median hourly wage for medical records specialists was $24.59.

Scope: The national estimate combines six survey panels collected over three years.

Table 1, PDF page 13, Medical records specialists row.

HCC Buddy recorded 133,342 strict ICD-10-CM code lookup events from April 11 through July 9, 2026.

Scope: Eligible events in a fixed 90-day window, not unique coders or patients.

originalData.statistics[0] in the JSON dataset.

From April 11 through July 9, 2026, HCC Buddy recorded strict ICD-10-CM lookup events across 1,316 official three-character code families.

Scope: Dots were removed before grouping, and each family was checked against the official fiscal year 2026 ICD-10-CM tabular.

originalData.statistics[1] in the JSON dataset.

E11 (Type 2 diabetes mellitus) was the leading code family with 5,602 events, or 4.2% of eligible lookups.

Scope: An aggregate lookup count from official fiscal year 2026 code families, not a prevalence estimate or patient count.

originalData.statistics[2] in the JSON dataset; label checked in the fiscal year 2026 CMS tabular and icd10-hcc MCP.

The ten leading ICD-10-CM families accounted for 25,619 events, or 19.2% of eligible lookups.

Scope: Based on the same fixed 90-day event window and official fiscal year 2026 family filter.

originalData.statistics[3] in the JSON dataset.