Performance Improvement 2008. How Appropriate Are Medicare Drug Coverage Risk-Adjustments for the Disabled and Low-Income Subsidy-Eligible Persons?


This project continued research on drug risk adjustment originally designed to supplement Federal Employee Health Benefit Program (FEHBP) data used to develop the initial drug risk adjusters. Drug risk-adjustment is the process by which payments to drug insurance plans are adjusted to take into account the higher costs of individuals who have expensive medical conditions which require greater use of pharmaceuticals in order to neutralize incentives for drug plans to enroll only healthy individuals with low drug utilization in order to maximize profits.  

There were known deficiencies in the FEHBP data. For example, the data set was not nationally representative, it had few Medicaid, disabled or low-income representatives and it reflected a relatively well insured population prior to age 65. Researchers matched Medicare and Medicaid data for dually eligible individuals for the same year the FEHBP data was available and looked at the utilization patterns for the low-income, institutionalized, and disabled. These patterns were used by Medicare actuaries to develop risk adjustment weights for low income and disabled populations.

Further efforts were made to merge the Medicaid data with the FEHBP data so that the drug prices in each database were on the same scale and so that direct calculations of the relative drug costs for each population subgroup could be made. The initial study of the geographic variation of drug utilization and price patterns by State was expanded to look at the drug regions established by HHS and other factors such as urban-rural areas of States. The study examined what could be learned from drug utilization data to improve a prospective diagnostic-based drug risk adjustment model. (For example, if only people with diabetics used insulin, how much could the risk adjustment model be improved if information on insulin use was available, but diabetes was not coded in the diagnostic data?)

The study yielded some important patterns which were unexpected, e.g., drug use was higher per capita in areas with fewer physicians. The data were not strong enough to warrant a policy change in the absence of Part D data. Finally, the prior use study concluded that additional data could improve the prospective risk adjustment model. Care would have to be taken that prior-use was not substituted for diagnostic data––a fine line that might be difficult to maintain.

Report Title: Continuation of Drug Risk Adjustment,
Agency Sponsor:
ASPE-OHP, Office of Health Policy
Federal Contact: George Greenberg, 202-690-7794
Performer: National Opinion Research Center (NORC)

View full report


"PerformanceImprovement2008.pdf" (pdf, 1.29Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®