Tag Archives: hierarchical condition categories

CMS proposed changes to the Medicare Advantage risk adjustment model

Late last month, the Centers for Medicare and Medicaid Services (CMS) released a 60-day “Advance Notice of Methodological Changes for Calendar Year (CY) 2019 for the Medicare Advantage (MA) CMS-HCC Risk Adjustment Model,” which describes proposed changes to the 2019 Part C risk adjustment model. CMS is seeking comments on the proposed changes, which are due by March 2, 2018.

The 21st Century Cures Act (Cures Act) requires CMS to make improvements to the CMS-HCC risk adjustment model for 2019 and subsequent years. The Cures Act directs CMS to:

• Evaluate the impact of including additional diagnoses for mental health and substance abuse disorders, as well as chronic kidney disease
• Make adjustments to the risk payments to account for the number of diseases or conditions of a beneficiary
• Phase-in the above changes to the risk adjustment payment over a three-year period, beginning with 2019 and fully implemented for 2022 and subsequent years

Based on the evaluation of the additional diagnosis codes, CMS is proposing to add to the model three new Hierarchical Condition Categories (HCCs) related to mental health and substance abuse, and one new HCC related to chronic kidney disease. In addition, CMS is proposing to include additional diagnosis codes for an existing substance abuse HCC.

In order to account for the number of conditions for each beneficiary, CMS has proposed to include new HCC count variables in the proposed risk adjustment model. As part of the development of the new count variables, CMS compared the predictive power of a model that counts only the conditions that result in a payment to MA plans in the CMS-HCC model (“Payment Condition Count” model) to a model that counts all conditions, regardless of whether they are used for risk payment (“All Condition Count” model). CMS concluded that the “Payment Condition Count” model increased the predictive accuracy of the risk adjustment model, while the “All Condition Count” model decreased the predictive accuracy. Both models are included in the Advance Notice for comment. CMS also noted that, in order for the overall fee-for-service (FFS) risk score to remain revenue-neutral, adding the new count variables would result in a decrease to the coefficients for many HCCs.

For 2019, CMS is proposing a model phase-in schedule that blends 25% of the risk score calculated using the proposed “Payment Condition Count” model and 75% of the risk score calculated using the existing 2017 CMS-HCC model. The weights of the “Payment Condition Count” model are proposed to increase to 50% in 2020, 75% in 2021 and 100% in 2022. However, CMS comments that because the three-year phase-in is required over a four-year period (2019 to 2022), it may be possible to use 2019 for comments and implement model changes in 2020.

In addition to the requirements directed by the Cures Act, CMS is proposing to recalibrate the 2019 CMS-HCC model using more recent data. The Advance Notice also proposes an increase to the weight given to the Encounter Data System (EDS) risk scores, from 15% in 2018 to 25% in 2019; these weights are used to blend the EDS and Risk Adjustment Processing System (RAPS) risk scores during the transition to 100% EDS. For 2019, CMS is proposing to combine the two phase-ins (increasing the weight for EDS and adding the proposed new model) by using the “Payment Condition Count” model exclusively for EDS risk scores and the existing 2017 CMS-HCC model exclusively for RAPS risk scores. Hence, CMS will only calculate two risk scores, one using the proposed model and EDS data at a 25% weight, and the second using the existing model and RAPS data at a 75% weight. CMS also plans to include RAPS inpatient submissions as an additional data source for the EDS risk scores, noting that inpatient submissions for EDS are low compared to RAPS. No explanation is offered for why inpatient submissions are low under the EDS methodology.

In the CMS fact sheet, it is stated that the new model will lead to an estimated 1.1% risk score increase across all MA plans, which equates to a 0.3% risk score increase after recognizing the 25% phase-in. However, results will vary for each plan. We expect to be able to evaluate the impact using actual data for individual plans once CMS releases the updated mapping of diagnoses to HCCs later this month.

The full text of the Advance Notice can be found here. The CMS Advance Notice Fact Sheet can be found here.

Medicare Advantage hierarchical condition categories: Updated study results

The pressure on Medicare Advantage (MA) plans to ensure that risk scores appropriately reflect the health status of their population under the Patient Protection and Affordable Care Act (PPACA) continues to increase. Payment rates from the fee-for-service (FFS) phase-in as well as changes in star ratings for MA plans have been impacted.

The Centers for Medicare & Medicaid Services (CMS) assigns a risk score to every MA member based on the member’s characteristics, including age, gender, disability status, Medicaid status, and “health” status. The majority of revenue received by MA plans is based on the risk scores of their members, and the health status is the primary variable in the calculation of the risk score.

CMS determines the diseases/hierarchical condition categories (HCCs) for each member based on ICD-9 diagnosis codes. Identifying and submitting all appropriate ICD-9 diagnosis codes to CMS results in a higher risk score for the member and an increased payment to the MA plan.

This article, first published in the October 2012 issue of the Society of Actuaries’ Health Watch newsletter, discusses accurate diagnostic coding as an important revenue tool.

Emphasis on quality in Medicare Advantage plans

We blogged last week about how payment reform is affecting Medicare Advantage. These plans are facing an increasing emphasis on quality and efficiency, both of which are complicated topics. For more on the quality dimension, consider this recent article from HealthWatch by Corey Berger and Eric Goetsch about hierarchical condition categories.