Costs and Savings under Federal Policy Approaches to Address Medicaid Prescription Drug Spending
Issue Brief
The National Health Expenditure Accounts report $20.9 billion in federal Medicaid prescription drug expenditures in 2019. Assuming spending is flat, the cost estimate translates to an 8% drop in federal Medicaid spending over the period. Assuming spending grows at the same average annual rate as the previous 10-year period, it translates to a 5% drop in federal Medicaid spending. Note that the policy would not be in effect for the entire budget period, since it does not go into effect until 2024.
MACPAC’s analysis finds $690 million per quarter in lost rebates; extrapolating this quarterly estimate leads to an annual estimate of $2.8 billion, on par with the CBO estimate of $3 billion in 2019.
Industry analysis of fee-for-service claims shows increases in net costs per claim for insulin from 2018-2019, indicating smaller rebates relative to gross spending at least on certain drugs in this class that could reflect these drugs continuing to hit the rebate cap.
This analysis, based on the State Drug Utilization Data, is based on spending per unit and only includes drugs that appeared in the data in both 2015 and 2019.
Among drug classes in which more than half of drugs increased faster than inflation, the median number of drugs was 7.5, compared to a median of 56.5 among classes in which less than half of drugs increased faster than inflation.
PhRMA, Comments Of The Pharmaceutical Research And Manufacturers Of America, (PhRMA 2018), https://www.phrma.org/-/media/Project/PhRMA/PhRMA-Org/PhRMA-Org/PDF/P-R/PhRMA-RFI-Comments-on-HHS-Blueprint-to-Lower-Drug-Prices-and-Reduce-Out-of-Pocket-Costs5.pdf
KFF analysis of 2019 State Drugs Utilization Data (SDUD) data, https://www.medicaid.gov/medicaid/prescription-drugs/state-drug-utilization-data/index.html.
The rebate amount for brand-name drugs is a defined percent of Average Manufacturer Price (AMP) or the difference between AMP and “best price,” whichever is greater. Best price only applies to brand drugs; generic drug rebates are calculated based on 13% of AMP.
The drugs included in the CBO analysis accounted for 62% of total spending on brand-name drugs in Medicaid in 2017. Congressional Budget Office, A Comparison of Brand-Name Drug Prices Among Selected Federal Programs (CBO, February 2021), https://www.cbo.gov/system/files/2021-02/56978-Drug-Prices.pdf Our analysis assumes that the average rebate amount also applies to the 38% of brand-name drugs not included in the CBO analysis.
That analysis, conducted by Edwin Park and Andrea Noda, calculated $5 billion in discounts due to best price. The difference in these estimates is related to the higher rebate percentage in the 2021 CBO report compared to the 2019 CBO report upon which the earlier estimates were based as well as growth in Medicaid spending on brand-name drugs between 2015 and 2017.
This estimate is based on calculating an average minimum rebate of $118 (23.1% of the average $509 AMP) compared to an average rebate of $211 due to best price (AMP minus best price), a difference of $93. Over the total number of brand-name prescriptions covered by Medicaid in 2017 (102 million), this average savings translates to more than $9.5 billion in additional discounts due to best price. However, the CBO analysis is based on standardized prescriptions, while the 102 million prescriptions are not standardized and are simple counts at the NDC level.
Based on analysis of 2019 State Drug Utilization Data (SDUD). We ranked drugs according to the cost per prescription; this analysis therefore does not reflect cost per course of treatment and is only illustrative of the scale of best price savings for very high cost drugs.
This rough calculation assumes an additional 4 percentage point rebate on approximately 35% of gross Medicaid drug spending.
This calculation is based on gross spending per prescription for drugs for which an NDC code newly-appears in the State Drug Utilization Data (SDUD) between 2011 and 2018. Because SDUD suppresses data for drugs with a small number of claims, and because many new NDCs have spending suppressed in the first year they appear, we calculated gross spending per prescription for the second year for which an NDC newly-appears in the data. Still, about a third of drugs for which an NDC newly-appears in the SDUD data have data suppressed even in subsequent years. We compared the gross spending per prescription for new NDCs to median gross spending per prescription for other drugs in the same therapeutic class that year. A small number of drugs (0.4%) did not have a comparison therapeutic class due to being the only drug in their class that year and were excluded.