The Association Between the Medicare Advantage Value-Based Insurance Design Model Test and Beneficiary Risk Scores

The enclosed memoranda document the results of the VBID model evaluation contractor's additional analyses. These analyses expand on the 2023 VBID evaluation report finding of an association in 2020 between VBID plan participation and higher beneficiary-level risk scores among enrollees eligible for VBID interventions. These additional analyses had two objectives:

1. Identify whether the overall association between the VBID model and higher enrollee risk scores was concentrated among specific VBID intervention(s) or group(s) of model participants (e.g., Dual Special Needs Plans (D-SNPs), non-SNPs); and
2. Assess which parts of the risk score calculation drove the 2023 evaluation report's observed increase in enrollee risk scores.

To the first objective, the results do not indicate that the overall risk score association with the model was attributable to specific VBID intervention(s) or group(s) of participants. Although the magnitude varied depending on the specific subset of the model that was analyzed (e.g., targeting of reduced or eliminated Part D cost-sharing), the association of the VBID model with higher risk scores among VBID-eligible enrollees persisted across the specific VBID interventions examined, and also was present for enrollees in both D-SNPs and non-SNPs that participated in the model.

To the second objective, the results of the additional analyses identified an increase in the number of Hierarchical Condition Categories (HCCs) for VBID-targeted enrollees, relative to comparable enrollees in plans that did not participate in the model. This included statistically significant increases in the prevalence of HCCs that were both related and unrelated to VBID interventions. An analysis of the rate at which continuously-enrolled enrollees became dual eligible, which is included in the risk score calculation, did not find a substantial association between increased enrollee dual eligibility and their plan's VBID participation status.

Based on these additional findings, the association between VBID plan participation and higher enrollee risk scores was seen across subsets of the model and that the increased prevalence of HCCs drove the risk score increase associated with the VBID model.