The Final 2022 Payment Notice: Risk Adjustment

On April 30, 2021, the Centers for Medicare and Medicaid Services (CMS) released the rest of the final 2022 Notice of Benefit and Payment Parameters Rule. The 473-page final rule was accompanied by a fact sheet and a press release. The “payment notice” is issued on an annual basis to adopt a variety of major changes that CMS intends to implement for the next plan year in areas such as the marketplaces, the risk adjustment program, and the market reforms. The proposed 2022 payment rule was released on November 25 with comments due on December 30. Federal officials received 542 comments on the proposed rule.

The Trump administration partially finalized the 2022 payment rule before inauguration day to adopt a subset of the policies considered in the proposed 2022 payment notice. This subset of policies included the proposed rule’s most controversial changes.

A first post addressed changes that generally apply to the exchanges, medical loss ratio (MLR) requirements, the coverage of essential health benefits, special enrollment periods, and reporting of prescription drug information by pharmacy benefit managers. This post considers the final rule’s changes to the risk adjustment program. CMS did not finalize proposed risk adjustment model specifications or enrollment duration factors; an option for states to request multi-year (instead of annual) reductions for risk adjustment transfers; and a shorter timeframe for insurers to confirm the findings of the second validation audit or the calculation of the risk score error rate.

Overview

The ACA included three premium stabilization programs: risk corridors, reinsurance, and risk adjustment. The risk corridor and reinsurance programs lasted from 2014 to 2016 while the risk adjustment program continues. The risk adjustment program transfers funds from lower-risk, non-grandfathered plans in the individual and small group markets to higher-risk, non-grandfathered plans, both in and out of the exchange. The purpose of the program is to discourage cherry picking: plans that end up with healthier populations must compensate plans that have more costly enrollees. CMS will operate the risk adjustment program in all states in 2022.

The risk adjustment model predicts plan liability for an average enrollee based on risk scores. These risk scores are based on each enrollee’s age, sex, and diagnoses. Diagnoses are also referred to as hierarchical condition categories (HCCs). The risk adjustment methodology uses separate models for adults, children, and infants to account for cost differences. In the adult and child models, each person’s age, sex, and diagnoses are added together to produce an individual risk score. For adults, CMS has added data over time, such as enrollment duration factors and prescription drug categories (RXCs).

Where applicable, risk scores are multiplied by a cost-sharing reduction adjustment, recognizing that enrollees with lower cost-sharing use more services. CMS has maintained the same cost-sharing reduction adjustments since 2019 and does the same for 2022. The enrollment-weighted average risk score of all enrollees in a particular risk adjustment-covered plan within a geographic area is an input into the risk adjustment payment transfer formula, which determines the payment or charge an insurer will receive or have to pay under the program.

Unlike, say, the risk corridor program, risk adjustment had been relatively low-profile. Parts of the risk adjustment methodology were challenged in court, leading CMS to briefly suspend risk adjustment transfers. But that litigation was resolved in late 2019 after an appellate court upheld the challenged part of the methodology.

Risk Adjustment User Fee

The final risk adjustment user fee for 2022 is $0.25 per member per month, unchanged from 2021. CMS estimates that it will cost about $60 million in benefit year 2022 to operate the risk adjustment program.

Recalibration

With the goal of improving the accuracy of its risk adjustment models, CMS transitioned to only External Data Gathering Environment (EDGE) data for recalibration beginning with the 2021 payment rule (although this transition had been a long time coming as CMS incorporated enrollee-level EDGE data over time). Going forward, CMS will use the three most recent consecutive years of enrollee-level EDGE data that are available to be included in annual proposed payment rule and not update the coefficients for additional years of data between the proposed and final rules even if an additional year of EDGE data becomes available.

For 2022, the available enrollee-level EDGE data is from 2016, 2017, and 2018; these are the same three years of data as used for the 2021 plan year. CMS did not incorporate data from 2017, 2018, and 2019 because the 2019 EDGE data was not yet available.

Comments were mixed. Some commenters agreed with the proposed approach so as not to delay final risk adjustment coefficients; others argued that newer data is more important than having the model coefficients be released earlier. CMS does not anticipate dramatic changes to risk scores for 2022 recalibration; the use of the same data will maintain year-to-year stability in risk scores. CMS clarifies that there could be other reasons why the coefficients change between future proposed and final payment rules, but it would not be because of newly released data.

Though CMS adopts the methodology for the risk adjustment coefficients, it is not finalizing the proposed coefficients because they reflected the proposed model updates which, as discussed below, were not finalized. The coefficients included in the final rule are the correct ones to use.

While not directly relevant to the 2022 payment rule, several commenters asked CMS about how it will consider enrollee-level EDGE data from 2020 given the impact of the pandemic. CMS intends to carefully review the 2020 EDGE data and consider whether this data should be included in recalibration for future years, or not. Because CMS will use data from 2016, 2017, and 2018 for the 2022 model recalibration process, federal officials and stakeholders have an additional year to evaluate the data from 2020 and assess the implications of using it for recalibration purposes. The earliest point at which 2020 EDGE data would be used for recalibration is for the 2024 benefit year.

Risk Adjustment Model Updates

CMS did not finalize proposed updates to its risk adjustment model specifications or enrollment duration factors but will further consider potential changes that could increase the predictive power of the risk adjustment models. CMS will continue Hepatitis C pricing adjustments.

Model Specifications

CMS has long acknowledged concerns that the risk adjustment models slightly under-predict risk for low-cost enrollees (those without HCCs) and slightly over-predict risk for high-cost enrollees (those with the highest HCC counts). The models also slightly over-predict for enrollees with low HCC counts. This discrepancy affects the risk scores of plans that enroll healthier individuals and plans that enroll the sickest individuals. To address this issue, CMS outlined various options—such as a constrained regression approach, an adjusted risk scores approach, and non-linear and count model specifications—in the 2021 payment rule but did not settle on a new approach.

In the proposed 2022 payment rule, CMS would have adopted a two-stage model specification in the adult and child models, added interacted HCC count factors, and removed severity illness indicators beginning with the 2022 benefit year. These technical changes are discussed at length in the preamble to the final rule.

But, after soliciting stakeholder comment, CMS opted not to adopt those changes. Instead, the current risk adjustment model specifications that were finalized in the 2021 payment notice will continue to apply, with trending adjustments made to project the data used to develop the factors forward to reflect the 2022 benefit year. This includes the current severity illness indicators in the adult models.

Some commenters supported the changes. Others raised concerns that CMS had not given enough information or time to assess the model specification questions. Some thought that CMS should first issue a white paper and conduct listening sessions. Still others believed that the changes would add complexity and hinder accurate pricing, resulting in higher premiums, and raised more technical concerns about how the model specification would operate. CMS believes stakeholders had sufficient time and information to consider the impact of the models but still decided not to finalize the proposed changes at this time. It may also issue a technical paper to better inform stakeholder understanding.

Enrollment Duration Factors

CMS proposed changes to enrollment duration factors in the adult model. Preliminary analysis of 2017 EDGE data showed that partial year enrollees with HCCs had higher expenditures on average compared to full-year enrollees with HCCs. At the same time, expenditures for partial year enrollees without HCCs did not differ compared to expenditures for full-year enrollees without HCCs. Enrollment timing (e.g., the beginning or end of the year) also did not affect expenditures. CMS reassessed enrollment duration factors using 2018 EDGE data and reached similar conclusions. It also found that current enrollment duration factors underpredict plan liability for partial year adult enrollees with HCCs and overpredict plan liability for partial year adult enrollees without HCCs.

In light of this data, CMS considered removing the current 11 enrollment duration factors of up to 11 months for all enrollees in the adult models and adding new monthly enrollment duration factors of up to 6 months only for enrollees with HCCs. This would have meant no enrollment duration factors for adults without HCCs beginning with the 2022 benefit year.

CMS did not, however, finalize this proposal. Instead, the current 11 enrollment duration factors of up to 11 months for all enrollees in the adult model (with trending adjustments), will continue to apply for 2022. CMS will continue to analyze potential changes to enrollment duration factors to improve model prediction for partial year enrollees with HCCs.

Many commenters opposed the this proposed change and asked for additional analysis and further evaluation in a white paper or through stakeholder listening sessions. Others raised substantive concerns, arguing that replacing current enrollment duration factors are appropriate and help mitigate potential under-prediction issues in the small group market. Some asked for alternatives, such as tying factors to only certain HCCs (such as maternity) or creating enrollment duration factors up to 9 months (rather than 6 months as proposed). CMS reviewed the data on these latter points and believed that accommodating those suggestions would add too much complexity and potential instability to the models or would not meaningfully improve predictive ratios.

Hepatitis C Pricing Adjustments

Consistent with prior years, CMS is concerned that the RXC for Hepatitis C does not fully account for pricing changes due to costly new Hepatitis C drugs and therefore does not accurately reflect the average cost of Hepatitis C treatments. CMS is also aware that insurers may try to influence providers’ prescribing patterns if a drug claim can trigger an increase in the enrollee’s risk score, resulting in a more favorable risk adjustment transfer.

To address these concerns and avoid perverse incentives to influence overprescribing behavior, CMS will continue applying a market pricing adjustment to plan liability associated with Hepatitis C drugs. Since CMS relies on the same three years of EDGE data as for 2021, the market pricing adjustment here will be prior to solving for coefficients for the 2022 benefit year models. This policy was first adopted in the 2020 benefit year risk adjustment models, and most commenters supported the continuation of this pricing adjustment. CMS intends to reassess this pricing adjustment using EDGE data in the future.

Beyond Hepatitis C treatments, some commenters asked that the risk adjustment methodology account for new expensive therapies and treatments (such as gene therapy drugs) and ancillary services associated with pre-exposure prophylaxis (PrEP). CMS responded by noting that the data used to recalibrate the risk adjustment model always lags by several years and that expensive new therapies can be addressed by high-cost risk pool payments. With respect to PrEP, costs are treated the same for PrEP as under the risk adjustment model and, for now, CMS treats ancillary services for PrEP as preventive services. CMS will continue to consider whether additional PrEP ancillary services should be treated as preventive services for purposes of risk adjustment in future benefit years.

Payment Transfer Formula

CMS will continue to use the state payment transfer formula that was finalized in the 2021 payment rule. Once it has calculated the risk scores for each plan’s enrollees, CMS feeds these into its payment transfer formula to determine, for each geographic area in a state, per member per month amounts to be transferred among plans as payments or charges based on each plan’s total member months for the plan year. CMS will not republish the formula in future payment notices unless changes are being proposed.

Transfers are based on the statewide average premium and reduced by 14 percent to account for administrative costs that do not vary with claims. CMS will exclude the costs of enrollees whose claims exceed $1 million when calculating enrollee-level plan liability risk scores. Plans will be compensated directly for 60 percent of costs above the $1 million threshold. These parameters are consistent with prior years.

Future Premium Credits

Consistent with changes to medical loss ratio requirements, CMS codifies its recent temporary policy of relaxed enforcement and interim final rule regarding temporary premium credits. Under this policy, insurers in the individual and small group markets could temporarily reduce monthly premiums for 2020 coverage so long as they meet certain requirements.

In the final 2022 payment rule, CMS laid out how it would treat temporary premium credits for purposes of risk adjustment in future benefit years if there is a declared public health emergency and when permitted by CMS. Consistent with the interim final rule, insurers with risk adjustment-covered plans would have to report the lower “adjusted plan premium” to the EDGE server. This refers to the premium amounts that are actually billed to enrollees (including any reduction in premiums because of a premium credit). The calculation of risk adjustment transfers, then, would be based on reported premiums and statewide average premiums that reflect adjusted plan premiums. These changes would not otherwise affect the state payment transfer formula or methodology.

State-Specific Adjustments

In the 2019 payment rule, CMS gave states the flexibility to request a reduction to their risk adjustment transfers by up to 50 percent of the premium used in the applicable plan year. The goal of state-specific adjustments is to allow for risk adjustment transfers that more precisely account for differences in the risk in a state’s market.

States must demonstrate that state-specific factors warrant an adjustment and that an adjustment would have a de minimis effect on premium increases. States must submit their requests and supporting data to CMS by August 1 of each year. CMS will publish whether state reduction requests were approved or denied and can approve a reduction that is lower than what a state requested, if warranted.

For 2020 and 2021, CMS approved requests from Alabama—the only state to make such a request—to reduce risk adjustment transfers in its small group market by 50 percent. Alabama made the same request for 2022 and was, again, the only state to do so. Alabama additionally asked to reduce transfers in its individual market by 50 percent. State regulators continue to believe the risk adjustment program is not working as precisely as it should because Alabama has a dominant carrier in its individual and small group markets. CMS asked for comment on this proposal and released additional documentation for public review and comment.

CMS approved Alabama’s request reductions for both the individual and small group markets. CMS acknowledges comments that argued that this request was not justified but notes that it reviewed additional supporting evidence and analysis that could not be made publicly available. This evidence, CMS found, was sufficient to justify Alabama’s request.

CMS did not finalize a proposal that would have allowed states to request adjustments for up to three years at a time (in lieu of annual requests and approvals). Comments were mixed: some supported multi-year flexibility while others raised concerns about undermining the effectiveness of the risk adjustment program. CMS agrees with commenters that there are concerns and barriers to multi-year state flexibility reduction requests in part because market conditions can change significantly over the course of three years. Although the final rule does not allow for multi-year requests, states can continue to submit annual requests to reduce risk adjustment calculations.

RADV

To ensure that risk adjustment transfers are accurate, risk adjustment data collected from insurers must be validated, first by an independent validation auditor retained by the insurer and then by CMS. The insurer provides the auditor with demographic, enrollment, and medical record documentation for a sample of enrollees selected by CMS. The final 2022 payment rule includes several RADV changes and complements RADV updates included in a recently finalized rule.  

RADV Schedule

CMS overhauls its timeline to collect RADV payments and charges in the same calendar year in which the RADV results are released. This schedule had been updated in the 2020 payment rule, but CMS now will return to the previous schedule. This switch is designed to accommodate stakeholder concerns about the complexity of the existing timeline, a potential conflict with state financial accounting requirements, and negative impacts on MLRs.

Beginning with the 2019 benefit year RADV, CMS will collect RADV payments and charges in the calendar year in which RADV results are released. The RADV summary report will be issued no later than early summer, and insurers must report those amounts in MLR reports that year. CMS will then collect and disburse RADV adjustments and default charges or allocations in the summer or fall of the year when RADV results are released.

In case clearer, here is an example for 2021 RADV payments and charges. The collection and disbursement of 2021 RADV payments and charges will begin in 2023. The summary report on 2021 RADV adjustments and default charges will be released by early summer 2023, and insurers will report those amounts in the 2022 MLR reporting year, which is due by July 31, 2023. CMS will then collect and disburse 2021 RADV payments and charges in the summer or fall of 2023.

CMS will publish the 2019 and 2020 RADV reports in early summer 2022. Insurers must account for any payments or charges (and risk adjustment transfers for 2021) in their 2021 MLR reports due July 31, 2022. Then, CMS will collect both the 2019 and 2020 RADV adjustments to transfers and disbursing payments in late summer or early fall of 2022.

The preamble to the final rule included a summary chart of the risk adjustment and RADV benefit years to include in MLR reports based on MLR reporting years.

RADV Exemptions

CMS codifies a series of previously established RADV exemptions for insurers. An exemption is allowed for an insurer who only offers coverage in the small group market and exits that market but still has enrollees through the next benefit year (i.e., “carry-over” coverage). That insurer will be treated as an exiting insurer and thus exempt from RADV for the benefit year when it has carry-over coverage. An exemption will also be available for sole insurers in a state risk pool. This is because there are no risk adjustment transfers in states with only one insurer in the market.

Conflict Of Interest Standards For IVA Entities

CMS prohibits entities that conduct initial validation audits (IVA) for RADV from doing so if the entity had a prior role regarding that insurer’s risk adjustment or EDGE server data submission process. To demonstrate that the entity is reasonably free from conflicts, the IVA entity cannot have had a role in advising the insurer or establishing internal controls related to risk adjustment or the EDGE server data submission process. For example, the IVA entity cannot serve as an insurer’s IVA entity and third-party administrator for these functions during the same benefit year.

Appeals And Validation

CMS clarifies requirements for RADV administrative appeals and disputes over the findings of a second validation audit. CMS previously established a three-level administrative appeals process for insurers to seek reconsideration of risk adjustment charges, payments, and user fee calculations.

CMS declined to, as proposed, shorten the timeframe that insurers have to confirm the findings of the second validation audit or the calculation of the risk score error rate. This change goes into effect beginning with the 2020 benefit year RADV. The goal of this change was to help CMS resolve as many issues as possible ahead of releasing the RADV summary report. However, commenters raised concerns that the proposed 15-day timeline (down from 30 days) would not provide adequate time to complete a thorough review.

Audits And Compliance Reviews

With few changes from the proposed rule, CMS bolsters its audit and compliance authority and requirements related to the reinsurance and risk adjustment programs. Even though the reinsurance program ended in 2016, CMS wants to clarify its authority after facing challenges in performing reinsurance-related audits from the 2014 benefit year. Although CMS primarily uses the RADV process to audit the risk adjustment program, CMS intends to begin audits to ensure the proper payment of high-cost risk pool payments and confirm compliance with other risk adjustment requirements.

The changes in this area are detailed and not fully summarized here. But the new standards generally explain the audit process, how to comply with an audit, and the consequences for failing to comply. This includes notice and conference requirements, timelines, and obligations for downstream and delegated entities. CMS will enforce audit requirements if needed by recouping unsubstantiated reinsurance or risk adjustment payments made to an insurer.

EDGE Discrepancy Materiality Threshold

Following submission of EDGE data by insurers, CMS issues a final EDGE server report and provides insurers with an opportunity to confirm that the data is accurate or identify any discrepancies. CMS then has a process to address incorrect EDGE data submissions that materially impact risk adjustment transfers. Currently, this materiality threshold is at least $10,000 or one percent of the total estimated transfer amount in the applicable state market risk pool, whichever is less.

In the final rule, CMS codifies a materiality threshold for EDGE discrepancies of at least $100,000 or one percent of the total estimated transfer amount in the applicable state market risk pool, whichever is less. If the discrepancy is non-material, CMS will take no action. Further, CMS will only take action to address material discrepancies that harm other insurers in the same state market risk pool. If the discrepancy negatively affects the insurer that submitted it (but not other insurers in the market), the submitting insurer must accept the consequences—even if there is a substantial monetary impact for that insurer.

This threshold will go into effect beginning with the 2020 benefit year. This change does not affect other reasons why CMS might correct an EDGE discrepancy (such as a mathematical error or incorrect methodology, among other reasons). Most commenters supported the proposal.