ModifiedMedicarePolicy CMS-9913-F

Amendments to HHS-operated Risk Adjustment Data Validation (HHS-RADV) methodology and application timing

Final rule amending HHS-RADV error estimation methodology and the timing of applying HHS-RADV results; affects issuers in HHS-operated risk adjustment, auditors, and EDGE server processes.

Policy Summary

PayerMedicare

PolicyAmendments to HHS-operated Risk Adjustment Data Validation (HHS-RADV) methodology and application timing

Policy CodePolicy CMS-9913-F

Change TypeMethodology and application timing refinements (Super HCCs, sliding scale, negative constraint, transition averaging)

Effective Date[Insert date 30 days after the date of publication in the Federal Register ]

Next Review Date

Key ActionIssuers must maintain EDGE servers, submit required masked demographic and claims data, run HHS software, and prepare for audit and transition actions including the combined 2019/2020 adjustment for 2020 transfers.

SourceLink

POLICY UPDATE CHANGES

Finalizes refinements to the HHS-RADV error rate calculation beginning with the 2019 benefit year, including Super HCC grouping, a sliding scale adjustment for outlier confidence intervals, and a negative failure rate constraint; transitions application of results to the audited benefit year using an unweighted average of 2019 and 2020 for 2020 transfers.

Affirms adoption and continuation of RXC pilot inclusion and sampling refinements (Neyman allocation, expanded strata) to improve consistency across 2019–2020 HHS-RADV operations.

2019benefit year when error rate refinements begin

2020concurrent application begins

Average error rate approach

500member-month exemption (≤)

30min EDGE HCCs to be outlier

85.2% - 98.1%super HCC stability estimate

HHS-RADV Criteria and Error Rate Methodology

HHS-RADV methodological and application criteria

Finalized criteria for error estimation and application timing:

Adjust HCC grouping methodology to aggregate HCCs that share adult-model HCC coefficient estimation groups into Super HCCs for error estimation beginning with the 2019 benefit year.

Apply a sliding scale adjustment to issuer Group Adjustment Factors (GAF) to smooth outlier adjustments for issuer group failure rates near confidence interval bounds (apply linear adjustment between z = ±1.645 and ±3.00).

Constrain negative error rate outlier calculations when an outlier issuer has a negative failure rate by calculating the GAF as the difference between the weighted group mean failure rate (if positive) and zero for the negative portion of the GAF.

Transition application timing from prospective to concurrent application of HHS-RADV results for non-exiting issuers by combining 2019 and 2020 HHS-RADV results using an unweighted average and applying to 2020 benefit year risk scores and transfers; concurrent application begins with the 2020 benefit year.

children":[]}

HHS-RADV failure rate calculation and application criteria

Criteria and methods for calculating failure rates and determining outliers:

Calculate national weighted mean and standard deviation of failure rates across all issuers and compare each issuer's group failure rates to these national metrics to determine outlier status.

Aggregate HCCs that share the same adult-model HCC coefficient estimation group into a Super HCC; compute freqEDGE_c and freqIVA_c as the sum of constituent HCC frequencies and then compute the Super HCC failure rate FR_c.

Sort Super HCCs by FR_c and assign sequentially to low, then medium, then high failure-rate groups until cumulative freqEDGE reaches approximately 1/3 and 2/3 of total freqEDGE; where inclusion of a final Super HCC crosses a 1/3 boundary, generate multiple grouping scenarios and choose the assignment closest to an exact one-third split.

Apply an outlier issuer's error rate to adjust that issuer's applicable benefit-year plan liability risk score, which then recalculates the state market average risk score and adjusts transfers accordingly; for non-exiting issuers during transition, apply the unweighted average of 2019 and 2020 error rates to adjust 2020 risk scores and transfers.

Super HCC failure rate grouping procedure

Method for grouping and sorting HCCs for HHS-RADV failure rate calculations:

Combine HCCs that share an HCC coefficient estimation group in the adult risk adjustment models into Super HCCs prior to failure rate calculation; Super HCCs apply to all HHS-RADV sample enrollees regardless of model.

Calculate aggregated freqEDGE_c and freqIVA_c for each Super HCC as the sum of constituent HCC freqEDGE_h and freqIVA_h, and compute the Super HCC failure rate FR_c accordingly.

Sort Super HCCs by FR_c from lowest to highest and allocate to low/medium/high groups by sequentially adding Super HCCs until group cumulative freqEDGE meets the 1/3 and 2/3 cutoffs; if the final inclusion crosses a cutoff, generate and evaluate multiple assignment scenarios and select the scenario closest to an exact one-third distribution.

HHS-RADV error rate calculation criteria

Methodology and operational criteria affecting issuer error rate calculations and ensuing risk score adjustments:

Apply a linear sliding scale adjustment to the Group Adjustment Factor (GAF) for issuer group failure rates with z-scores between ±1.645 (90% CI) and ±3.00 (99.7% CI), such that issuers nearer the CI edge receive a graduated discount; issuers beyond ±3.00 receive full adjustment.

Allocate HCCs (or Super HCCs) into low, medium, and high failure rate groups and iteratively reallocate near-boundary Super HCCs to approximate equal one-third splits of total freqEDGE across groups per the specified reexamination procedure.

For negative error rate outliers with negative failure rates, constrain the GAF by substituting the weighted group mean with zero for the negative portion of the calculation, limiting upward adjustments driven by found HCCs.

These methodological changes are effective beginning with the 2019 benefit year and are not applied retroactively to 2017/2018 benefit years.

HHS-RADV error rate constraint criteria

Applicability and effect on HHS-RADV error rate calculation:

For negative error rate outlier issuers with negative failure rates, the GAF calculation is constrained such that the negative failure rate portion is calculated as the difference between the weighted group mean (if positive) and zero, thereby preventing credit for excessive 'found' HCCs beyond EDGE reporting in the negative portion of the GAF.

This negative failure rate constraint is integrated with the Super HCC grouping and sliding scale adjustments as part of the finalized error rate methodology beginning with the 2019 benefit year.

Found HCCs continue to be included in the national mean and failure rate calculations; the constraint only limits credit for found HCCs within the negative failure portion of a negative outlier's GAF.

Issuers with fewer than 30 EDGE HCCs within an HCC failure rate group will not be considered outliers for that group, though they remain included in national metric calculations.

HHS-RADV application and calculation criteria

How HHS-RADV results will be applied to adjust risk scores and transfers during transition and thereafter.

Transition rule: For the 2020 benefit year, non-exiting issuers' HHS-RADV adjustments will use an unweighted average of the 2019 and 2020 error rates (average error rate approach) to adjust 2020 plan liability risk scores and state market transfers.

Post-transition rule: Beginning with the 2021 benefit year, HHS-RADV results will be applied concurrently to the same benefit year being audited (for example, 2021 HHS-RADV results will adjust 2021 risk scores and transfers).

Exiting issuers: Exiting issuers remain subject only to positive error rate outlier application to the benefit year being audited; the unweighted average transition approach does not apply to exiting issuers.

New entrants: New market entrants are not exempt from HHS-RADV; after the transition, new entrants will only be affected by HHS-RADV results for the benefit year(s) in which they participated.

Error rate calculation components: Applied error rate calculations incorporate Group Adjustment Factors (GAF), Super HCC aggregation, sliding scale adjustments, the negative failure rate constraint, and extrapolation of amended enrollee EDGE risk scores across sample strata per the finalized formulas and procedures.

HHS-RADV application and methodology criteria

Operational criteria for applying HHS-RADV results during transition and methodological application:

Concurrent application of HHS-RADV results begins with the 2020 benefit year; non-exiting issuers' 2019 and 2020 results will be combined via an unweighted average to adjust 2020 risk scores and transfers during the transition.

Continue the RXC pilot for 2020 to increase consistency between 2019 and 2020 HHS-RADV operations and sampling; issuers should maintain EDGE submissions and IVA/SVA processes in anticipation of concurrent application.

Finalize methodological refinements to error estimation: Super HCC grouping based on adult-model coefficient estimation groups, sliding scale application near CI bounds, and constraints on negative failure rates as part of the revised error rate methodology.

Monitoring/contingency: HHS will monitor impacts (including COVID-19-related effects) on future benefit years and may consider additional flexibilities if warranted.

HCCs, Super HCCs, RXCs and Related Coding Rules

Drug-derived diagnosis categoriesmixed

RXC	Risk adjustment prescription drug categories to impute diagnoses or severity in hybrid model

HCC coefficient estimation groups (adult model V05)mixed

HHS-HCC coefficient estimation groups (e.g., G01..G18 etc.)

Lists of adult-model HCC coefficient estimation groups (V05) that are candidates for aggregation into Super HCCs.

Super HCC grouping keymixed

HCC coefficient estimation groups

Grouping key used to form Super HCCs by aggregating HCCs that share adult-model coefficient estimation groups for failure rate calculation (freqEDGE_c, freqIVA_c, FR_c).

HCC / Super HCCmixed

HCC	Hierarchical Condition Category codes subject to grouping into Super HCCs and to application of Group Adjustment Factors (GAF) applied to enrollee EDGE HCC coefficients as part of adjusted enrollee risk score calculation.

HCC grouping refinementsmixed

HCC grouping

Refinement to group HCCs that share an adult-model coefficient estimation group into Super HCCs for failure rate grouping; iterative allocation to approximate one-third freqEDGE splits and preserve risk differences within hierarchies.

Materiality/member month exemption

Exemption criteriaIssuers with 500 or fewer billable member months are exempt from annual IVA (materiality/member-month exemption); issuers with total annual premiums at or below $15 million are subject to a triennial IVA cadence instead of annual IVA.

Effective scopeMateriality/member-month exemptions reflected in HHS-RADV procedures and timelines as codified in the 2020 Payment Notice and prior guidance.

Operational implicationExempt issuers remain included in national metric calculations though may not undergo annual IVA audits; exemption reduces issuer audit burden.

Outlier HCC count floor

Issuer Operational Responsibilities and Actions

Note

Issuer data submission and audit responsibilities

Issuers must operate and maintain EDGE servers and execute HHS-provided software to generate the summary reports and data HHS uses to calculate enrollee-level and plan liability risk scores. Issuers remain responsible for the accuracy and completeness of data submitted to EDGE and for coordinating with IVA and SVA entities per 45 CFR 153.630. HHS may validate a statistically valid sample of issuer-submitted risk adjustment data (HHS-RADV) and adjust plan average actuarial risk and transfers based on HHS-RADV findings per 45 CFR 153.350. Issuers should anticipate that HHS-RADV findings can affect their own transfers and, because the program is budget neutral, transfers for other issuers in the same state market risk pool.

Maintain an EDGE server and run HHS-provided software to generate required summary reports.
Engage independent IVA auditors and provide demographic, enrollment, claims, and medical records for HHS-selected samples; cooperate with SVA as applicable.
Expect HHS-RADV to be applied to adjust plan liability risk scores and pool transfers; adjustments to one issuer affect all issuers in the same state market risk pool due to budget neutrality.

Key Terms and Acronyms

EDGE server definition

DefinitionEDGE server: External Data Gathering Environment (EDGE) server where issuers submit masked enrollee demographics, claims, and encounter-level diagnosis data and run HHS-provided software to generate summary reports.

Issuer responsibilityIssuers must establish and maintain EDGE servers in the format specified by HHS and execute HHS software to produce reports used for enrollee-level risk score calculations.

UseEDGE outputs are used to compute enrollee risk scores, state market average scores, and to support HHS-RADV failure rate and error rate calculations.

IVA and SVA

IVAInitial Validation Audit (IVA): issuer-engaged independent auditors perform IVA on HHS-selected samples to validate issuer-submitted EDGE data.

SVA

Policy Changes and Effective Dates

2019 (benefit year)methodology_effective

Finalized methodological refinements to HHS-RADV error rate calculation (including Super HCC grouping, sliding scale adjustment, and negative failure rate constraint) to begin applying with the 2019 benefit year.

2020 (benefit year)application_transitionLatest

Transitioned application timing: HHS will apply HHS-RADV results to the audited benefit year using an unweighted average of 2019 and 2020 error rates to adjust 2020 benefit year plan liability risk scores and transfers for non-exiting issuers.

Policy Summary

PayerMedicare

PolicyAmendments to HHS-operated Risk Adjustment Data Validation (HHS-RADV) methodology and application timing

Policy CodePolicy CMS-9913-F

Change TypeMethodology and application timing refinements (Super HCCs, sliding scale, negative constraint, transition averaging)

Effective Date[Insert date 30 days after the date of publication in the Federal Register ]

Next Review Date

SourceLink

On This Page

Recognize that Super HCC grouping is expected to preserve assignment for the majority of EDGE occurrences (analysis showed ~85–98% stability across 2017–2018 test data) while improving stability for less common conditions.

Issuers with fewer than 30 EDGE HCCs within an HCC failure rate group will not be determined to be outliers for that group (minimum EDGE HCC threshold = 30).

Inclusion in national metricsIssuers below the 30-HCC floor are still included in the calculation of national weighted mean and standard deviation for failure-rate groups.

Effective benefit yearFinalized and effective beginning with the 2019 benefit year (as codified in the 2021 Payment Notice clarifications).

HCC failure rate group assignment

Assignment ruleRank HCCs by national average failure rate and divide into three failure-rate groups (low/medium/high) such that each group's total freqEDGE is approximately one-third of total EDGE HCC frequency.

Super HCC aggregationPrior to grouping, aggregate HCCs that share an adult-model HCC coefficient estimation group into a single 'Super HCC' and compute freqEDGE and freqIVA for each Super HCC for ranking.

Iterative allocationIf adding a Super HCC causes a group to exceed the 1/3 cutoff, reexamine alternative assignments and choose the scenario closest to an exact one-third split.

Failure rate group cutoffs

1/3 cutoffsSort Super HCCs by failure rate and sequentially assign to low, then medium, then high until cumulative freqEDGE reaches ~1/3 and ~2/3 thresholds; aim for equal total freqEDGE across groups.

Boundary reexaminationWhen inclusion of the final Super HCC causes a group to exceed the 1/3 boundary, generate grouping scenarios by moving crossing Super HCCs into the next group and select the assignment with the smallest absolute distance from exact one-third splits.

Protocol referenceProcess follows HHS-RADV sorting rules and will be documented in the applicable HHS-RADV Protocols for the benefit year.

Sliding scale z-score thresholds

Sliding scale z-score thresholdsInner z-score (innerZ) = ±1.645 (90% CI); Outer z-score (outerZ) = ±3.00 (99.7% CI). The sliding scale is applied linearly between these bounds.

Confidence interval coverageSliding scale adjustment covers issuer group failure rates between the 90% and 99.7% confidence interval bounds to smooth payment-cliff effects.

Full adjustment beyond boundsFor issuers with z-scores beyond ±3.00, the GAF is fully applied without sliding-scale discounting.

Minimum EDGE HCCs to be considered outlier in group

Minimum EDGE HCCs per groupAn issuer must have at least 30 EDGE HCCs in a failure-rate group to be considered an outlier in that group (minimum = 30).

Operational noteIf an issuer does not meet the 30-HCC minimum for a group, it will not be identified as an outlier in that group but remains part of national metric calculations.

Effective dateApplied beginning with the 2019 benefit year as finalized in the 2021 Payment Notice updates and related rulemaking.

Failure rate grouping thresholds (examples shown)

Example thresholds (Low)Low group example (new grouping & 90% CI): weighted mean ~0.0369 (2017 data), lower/upper thresholds approx -0.1038 to 0.1777 (90% CI examples in Tables 3–4).

Example thresholds (Medium)Medium group example (new grouping & 90% CI): weighted mean ~0.1225 (2017 data), upper threshold approx 0.2633 (90% CI example).

Example thresholds (High)High group example (new grouping & 90% CI): weighted mean ~0.2595 (2017 data), lower threshold approx 0.0779 (90% CI example).

sliding scale confidence interval thresholds

Sliding scale CI boundsSliding scale applied between 90% and 99.7% confidence intervals (z = ±1.645 to ±3.00); an alternative using 95% to 99.7% was considered but not adopted.

Slope/intercept parametersThe sliding scale uses a linear slope (a) and intercept (b_r) defined in the finalized formulas to discount z-scores within the inner and outer z bounds.

Policy rationaleSelected range balances mitigation of payment-cliff effects while preserving HHS-RADV impact; alternatives (e.g., 1.96 innerZ) were evaluated but rejected due to weakening program impact.

Application of Error Rates to Risk Scores

HHS applies the outlier issuer's error rate to adjust that issuer's applicable benefit year plan liability risk score; that change affects the state market average and thus the transfers for the entire state market risk pool. To reduce volatility and improve predictability, CMS finalized refinements to the error rate methodology for HHS-RADV beginning with the 2019 benefit year, including combining certain HCCs into Super HCCs for failure rate grouping and constraining negative-error-rate outliers when failure rates are negative.

An outlier issuer’s error rate is applied to that issuer’s plan liability risk score and used to recalculate state market average risk scores, which changes transfers for the pool.
HCCs that share coefficient estimation groups in the adult models will be combined into Super HCCs for failure rate sorting beginning with the 2019 benefit year.
Negative failure rate constraint: negative-error-rate outliers with negative failure rates will have constrained impact on their General Adjustment Factor (GAF) to limit downstream effects on other issuers.

Note

Operational effect on issuers

Operationally, issuers should expect changes to timing and the way HHS-RADV results affect transfers during the transition to concurrent application of audit results. Beginning with the 2020 benefit year, CMS applies HHS-RADV audit findings to the benefit year being audited for all issuers by combining 2019 and 2020 results for non-exiting issuers using an average error rate approach; beginning with the 2021 benefit year, adjustments will be applied only once based on that same benefit year's HHS-RADV results.

Transition approach for 2019–2020: non-exiting issuers’ results combined (unweighted average error rate) to adjust 2020 transfers.
Starting with the 2021 benefit year, plan liability risk scores and transfers will be adjusted once based on the same benefit year's HHS-RADV results.
Exiting issuers will continue to be adjusted only for positive error rate outliers (not negative outliers) during the transition period.

Documentation Required

Access to HHS-RADV results

Issuers must ensure their internal processes and EDGE submissions align with HHS expectations because HHS-RADV results (including summary and issuer-specific Audit Tool results published for prior years) may be used to adjust transfers and will be shared publicly as summary information. Issuers should anticipate receiving issuer-specific and enrollee-specific audit results in the Audit Tool when HHS releases summary results.

Review prior-year HHS-RADV summary results and issuer-specific Audit Tool outputs to understand potential impacts.
Update internal documentation workflows to ensure medical record retrieval and response to IVA/SVA requests, especially when audits span multiple benefit years.
Monitor HHS postings (e.g., Premium Stabilization Program page) for published summary results and guidance.

Documentation Required

Issuer actions on EDGE submissions and internal processes

Issuers should update EDGE submission processes and internal controls to reflect the HCC grouping and error rate methodology changes, and to respond to IVA/SVA findings. Specifically, issuers should: verify coding and documentation practices to reduce missing HCCs; review reconciliation procedures between EDGE submissions and internal records; and ensure IVA entities have complete, accurately masked data and medical records for sampled enrollees.

Reconcile submitted EDGE data with internal claims and coding records before submission; prioritize accuracy for HCCs that may be aggregated into Super HCCs.
Strengthen documentation retrieval and tracking processes to support IVA and SVA audits and to reduce the likelihood of findings that would change reported HCCs.
Maintain audit trails for changes to EDGE submissions and document communications with IVA/SVA entities.

Note

Application and transition of HHS-RADV results

As HHS transitions to concurrent application of HHS-RADV results, issuers should plan operational changes to forecasting, reserve setting, and internal actuarial processes. Combine HHS-RADV results for transition years per CMS guidance (average error rate for 2019–2020) and prepare for single-year application beginning 2021. Expect constrained treatment of negative-error-rate outliers and the Super HCC grouping to affect the distribution of failure rates and resulting adjustments.

Incorporate the average error rate combination for transition years (2019 and 2020) into financial forecasting and reserves for non-exiting issuers.
Adjust actuarial models and reconciliation procedures to accommodate Super HCC groupings and constrained negative-error-rate calculations.
Plan for operational timing changes: audits, IVA/SVA coordination, and potential double adjustments during the transition if alternative timelines had been used (CMS finalized the average error rate approach).

Prior Authorization

Issuer operational actions for HHS-RADV transition

Immediate operational steps issuers should take for the HHS-RADV transition include updating policies, training relevant staff (claims, coding, medical records, actuarial), coordinating with IVA vendors to ensure sample handling matches HHS protocols, and revising financial models to reflect the concurrent application of HHS-RADV results starting with the 2020 benefit year and single-year application beginning in 2021.

Train operations, coding, and records teams on Super HCC groupings and implications for documentation and coding.
Engage IVA vendors early to confirm processes for sample delivery, documentation review, and timely responses to SVA.
Revise financial reporting and reserves to reflect combined 2019–2020 average error rate adjustments for transition and single-year adjustments thereafter.

Second Validation Audit (SVA): a subsample validated by an HHS-retained entity to verify IVA findings; used to confirm IVA accuracy and for potential substitution when IVA/SVA agreement is insufficient.

Role in error estimationIVA and SVA results feed into failure-rate calculations and extrapolated error rates used to adjust issuer plan liability risk scores.

RXC

RXC definitionRXC: Risk adjustment prescription drug category used to impute diagnoses or indicate severity in a hybrid diagnoses-and-drugs risk adjustment model; piloted in HHS-RADV beginning with 2018 and continued into 2019/2020 operations.

Operational noteRXC pilot extended for 2020 to increase consistency across benefit years and sampling; some sampled strata include only RXCs rather than EDGE HCCs.

HCC

HCCHCC (Hierarchical Condition Category): diagnostic groupings used in risk adjustment and subject to failure-rate grouping for HHS-RADV; HCCs are ranked by national failure rate and grouped into low/medium/high for outlier determination.

Role in enrollee risk scoreEDGE HCCs contribute HCC-specific risk score components that are adjusted via Group Adjustment Factors (GAF) when an issuer is an outlier in a failure-rate group.

Super HCC

Super HCC definitionSuper HCC: an aggregation of HCCs that share an HCC coefficient estimation group in the adult risk adjustment models; treated as one unit when computing failure rates and grouping for HHS-RADV.

Aggregation ruleAggregate freqEDGE and freqIVA across constituent HCCs to compute a Super HCC's failure rate (FR_c = 1 - freqIVA_c / freqEDGE_c).

ApplicabilitySuper HCCs are formed based on adult-model coefficient estimation groups and apply to all sampled enrollees regardless of which model (adult/child/infant) they are subject to.

Outlier issuer

Outlier issuerAn outlier issuer is one whose group failure rate for a given HCC failure-rate group falls outside the two-tailed 90% confidence interval (z = ±1.645) and meets the minimum HCC frequency requirement, triggering calculation of an issuer-specific error rate.

EffectAn outlier issuer's error rate is applied to adjust that issuer's applicable benefit year plan liability risk score, which in turn recalculates state market average risk scores and transfers.

Super HCC (alternate reference)

Alternate Super HCC referenceSuper HCC: formed by aggregating HCCs that share adult-model coefficient estimation groups; used in calculation examples (e.g., HCC 54 & 55 aggregation) to illustrate freqEDGE_c, freqIVA_c, and FR_c computations.

Example calculationExample: freqEDGE_c = freqEDGE_h54 + freqEDGE_h55; freqIVA_c = freqIVA_h54 + freqIVA_h55; FR_c = 1 - (freqIVA_c / freqEDGE_c).

freqEDGE / freqIVA / FR

freqEDGE / freqIVA definitionfreqEDGE and freqIVA are aggregated national frequencies from EDGE and IVA used to compute Super HCC failure rates (FR) prior to sorting into failure-rate groups.

Use in FR calculationFR_c is computed from aggregated freqEDGE_c and freqIVA_c for each Super HCC and used to rank and assign Super HCCs into low/medium/high groups.

Super HCCs / HCC grouping

Super HCCs / groupingSuper HCCs replace individual HCCs in the failure-rate grouping process; grouping is iteratively adjusted to approximate equal one-third allocations of total freqEDGE across low/medium/high groups.

RationaleFocused aggregation on adult-model coefficient estimation groups mitigates misalignment between model groupings and HHS-RADV failure-rate groups while preserving meaningful risk differences.

Sliding scale adjustment

Sliding scale adjustmentA linear discount applied to z-scores for group failure rates between innerZ = ±1.645 and outerZ = ±3.00 to smooth the payment-cliff effect; fully applied adjustments beyond ±3.00.

Implementation noteThe sliding scale modifies the Group Adjustment Factor (GAF) calculation by applying slope (a) and intercept (b_r) parameters to discounted z-scores within the selected range.

Policy goalMitigate abrupt payment-cliff changes for issuers near confidence interval bounds while retaining incentive effects for substantial deviations.

Negative failure rate constraint

Negative failure rate constraintFor negative error rate outliers with negative failure rates, constrain the GAF calculation by substituting the weighted group mean with zero, effectively calculating GAF as the difference between a positive weighted mean (if any) and zero to limit upward adjustments driven by found HCCs.

PurposeLimit credit for 'found' HCCs in negative failure cases so that large upward adjustments are not driven by previously unreported HCCs rather than genuine superior coding.

Effective startConstraint applies beginning with the 2019 benefit year as part of the finalized error rate methodology.

Super HCC (definition restated)

Super HCC restatedSuper HCC: aggregated HCCs sharing an adult-model coefficient estimation group; used as the unit for computing FR_c and for grouping into failure-rate groups in HHS-RADV.

ApplicationSuper HCC groupings apply to all HHS-RADV sampled enrollees regardless of the risk-adjustment model (adult/child/infant) to which they are subject.

GAF (Group Adjustment Factor)

GAF definitionGAF (Group Adjustment Factor): the adjustment factor applied to HCC failure-rate groups for an issuer, calculated based on issuer group failure rates, discounted via the sliding scale when applicable, and constrained for negative failure-rate outliers.

ApplicationGAF is applied to individual enrollee EDGE HCC coefficients to compute amended enrollee risk scores used in extrapolating issuer error rates.

Outlier determination

Outlier determinationAn issuer is an outlier in a given HCC failure-rate group if its group failure rate falls outside the two-tailed 90% confidence interval (z = ±1.645) and meets the minimum HCC count; sliding-scale and z-score based discounts then adjust the GAF magnitude.

Per-group independenceOutlier status is assessed independently for each failure-rate group, so an issuer may be an outlier in one group but not others.

GAF (alternate ref)

GAF alternate refGAF: calculated using z-scores, sliding-scale parameters (innerZ/outerZ, slope a, intercept b_r), weighted national mean and standard deviation of group failure rates, and constrained for negative failure-rate outliers as finalized.

Formula componentsKey components include z_G,i, disZ_G,i,r (discounted z-score), µ{GFR_G} (weighted mean), and Sd{GFR_G} (weighted standard deviation).

Adjusted EDGE risk score / enrollee adjustment

Adjusted EDGE risk scoreAdjusted EDGE risk score: enrollee-level amended risk score computed by applying the issuer's GAF to EDGE HCC coefficients for sampled enrollees with EDGE HCCs, used to extrapolate issuer error rates.

Extrapolation useDifferences between amended and original EDGE risk scores are extrapolated across strata to compute the issuer-level error rate applied to plan liability risk scores.

HHS-RADV error rate

HHS-RADV error rateError rate: the extrapolated difference between amended and original EDGE risk scores across sample strata for an outlier issuer, applied to adjust that issuer's applicable benefit-year plan liability risk score per 45 CFR 153.350(b).

Application effectApplied error rates change the issuer's plan liability risk score and thereby recalculates state market average risk scores and transfers for the applicable benefit year.

Super HCC / HCC grouping (additional ref)

Additional Super HCC grouping noteHCC grouping refinement focuses on HCCs that share adult-model coefficient estimation groups rather than broader hierarchy aggregation to preserve risk-differentiation and documentation-difficulty distinctions.

Considered alternativesCMS analyzed broader hierarchy-based aggregation but finalized the coefficient-estimation-group-based Super HCC approach to better account for risk differences.

Sliding scale adjustment (definition restated)

Sliding scale definition restatedSliding scale adjustment: a graduated linear discount applied to GAF calculations for issuer group failure rates between innerZ = ±1.645 and outerZ = ±3.00 to reduce payment-cliff impacts while retaining full adjustments beyond ±3.00.

Alternatives consideredCMS considered alternative innerZ choices (e.g., ±1.96) and ranges (95%–99.7% CI) but selected ±1.645 to balance program impact and smoothing goals.

Negative failure rate constraint (restated)

Constraint summaryNegative failure rate constraint: when an outlier issuer has a negative failure rate, the GAF calculation substitutes the weighted group mean with zero so that the GAF equals the weighted mean (if positive) minus zero, limiting upward adjustments driven by found HCCs.

Policy intentPrevent excessive upward adjustments that arise from newly found HCCs rather than true superior EDGE reporting; found HCCs still contribute to national means but are limited in negative-outlier credit.