MFN Drug Pricing Update: After GENEROUS, GUARD AND GLOBE Issue From CMS’s Innovation Center – Part I

In November, we posted that the New Year would be bringing us GENEROUS, a new Medicaid demonstration model initiated through CMS’s Center for Medicare and Medicaid Innovation (CMMI), which will permit drug manufacturers to enter into voluntary agreements with state Medicaid programs to provide most favored nation (MFN) pricing.  On December 23, CMS issued proposed regulations to establish, again under CMMI, two more MFN demonstration models under Medicare Parts B and D, respectively.  However, unlike GENEROUS, manufacturer participation in these models is mandatory.  The Part D model, called Guarding U.S. Medicare Against Rising Drug Costs (GUARD) is proposed to go into effect on January 1, 2027, while the Part B model, Global Benchmark for Efficient Drug Pricing (GLOBE), would become effective in the fourth quarter of 2026.  In this post, we outline the prominent features of the GUARD Model.  Tomorrow, we will post Part 2 of this article, which will focus on the GLOBE Model.  Comments on both the GUARD and the GLOBE proposals are due no later than February 23, 2026.

Duration:  GUARD would consist of five performance years beginning with 2027, for which manufacturers would be obligated to pay GUARD rebates, followed by a two-year period during which CMS would calculate, invoice and reconcile rebates owed for previous performance years.

Drugs Covered:  Drugs covered under GUARD would be sole source drugs and biologics that are subject to Medicare Part D inflation rebates and are within certain specified drug classes in the USP Medicare Model Formulary Guidelines.  “Sole source drugs” are defined as (1) single source drugs, meaning drugs approved under a new drug application (NDA), including authorized generics, that have no therapeutic equivalents listed in FDA’s Orange Book; and (2) biologics approved under a biologic license application (BLA), including unbranded biologics, that are not a reference biologic for a biosimilar application.  GUARD drugs that become multi-source during a performance year would no longer be subject to GUARD after that point.  There are 18 included USP drug classes, which are listed here.  They include the six Medicare protected classes and 12 additional classes.  Excluded would be generics approved under an ANDA, biosimilars licensed under section 351(k) of the Public Health Service Act, drugs selected for Medicare Maximum Fair Price (MFP) negotiation while the MFP is in effect, and drugs that do not meet a minimum Part D spend threshold, which is $69 million for 2027 and would be adjusted for inflation each performance year thereafter.

Geographical area and Medicare population covered:  No later than 60 days before each performance year, CMS would randomly select 25% of the Zip Code Tabulation Areas (ZCTAs) in the U.S.  ZCTAs are geographical areas roughly equivalent to zip codes.  Part D enrollees who reside in the selected ZCTAs (which would also represent approximately 25% of all Part D enrollees) would be covered under GUARD for that performance year.

Manufacturer participants:  Manufacturers subject to GUARD would be those that market GUARD drugs and are subject to the Medicare Part D inflation rebate program.  Participation is mandatory, and no enrollment procedure is necessary.

Basis for rebates – the International Pricing Benchmark:  In broad terms, the GUARD per-unit rebate for an NDC-9 in a performance year would be the excess (if any) of the Medicare Net Price over an International Pricing Benchmark, reduced by any Part D inflation rebate payable by the manufacturer for the same NDC-9.  Thus, the starting point for determining GUARD rebates is the International Pricing Benchmark.

The Benchmark would look at prices in the 19 foreign countries listed here.  These are members of the Organization for Economic Co-operation and Development (OECD) that have a purchasing power parity-adjusted per-capita gross domestic product (GDP) of at least 60% that of the U.S., and that have an annual GDP of at least $400 billion.

Benchmark prices would be developed for international analogs of a U.S. NDC-9, which would consist of foreign products with the same active pharmaceutical ingredient, dosage form, route of administration, and strength, excluding foreign generics and biosimilars.  Special procedures would be used to determine analogs where U.S. and foreign strengths are not aligned.

The International Pricing Benchmark for an NDC-9 would be the greater of two alternative benchmarks:  a “Default” Benchmark or an “Updated” Benchmark.  Under the Default Benchmark, CMS would use foreign pricing data obtained for the 12-months prior to a performance year (or, if unavailable, the most recent post-2024 12-month period available).  The data sources would be those that contain, in order of preference, sales plus volume (i.e., unit) data; price plus volume data; or prices only.  The preamble identifies three such sources:  IQVIA’s MIDAS, Eversana’s NAVLIN, and the Global Data Pharmaceutical Prices (POLI).

To determine the Default Benchmark for an NDC-9, CMS would calculate for each country, depending on the available data, either a volume weighted average price or a straight average price for the NDC-9 analog. To obtain the volume weighted average price, each different price for the analog in the country would be multiplied by its corresponding volume (in NCPDP units) to obtain a weighted price.  The weighted prices would then be added together and divided by the sum of the units for all prices.  For example, for an analog of an NDC-9 sold at three different prices in a country:

Volume weighted average price = [(price¹ x volume¹) + (price² x volume²) + (price³ x volume³)] ÷ [(volume¹ + volume² + volume³)]

If volume data were unavailable for a country, CMS would calculate a straight average price of the analog per NCPDP unit.  For example, for an analog sold in a country at three prices: (price¹ + price² + price³) ÷ 3.  The Default International Benchmark would be the lowest country average.

The second alternative method for determining the International Pricing Benchmark – the Updated Benchmark – would be derived from foreign pricing data voluntarily submitted to CMS by manufacturers, should they choose to do so.  A manufacturer that wanted to submit these data for any performance year would have to do so within 180 days after the end of the performance year.  The submission could cover one or more GUARD drugs, and would have to include data for all analogs (i.e., same API, dosage form, route of administration, and strength) of each GUARD drug sold in all 19 countries.  In addition to drug data (e.g., brand and nonproprietary name, approval status, API, route of administration, dosage form, strength, package sizes), the manufacturer would provide pricing data under either a “streamlined option” or a “limited option”.  The streamlined option would require, for each NDC-9 analog, all of the following:

• For each different net price in each country: gross sales, net sales, and sales volume in NCPDP units.
• For each country: (1) the volume weighted average net price; (2) the average net-to-gross ratio (total net sales ÷ total gross sales); (3) the exchange rate from the World Bank Atlas; and (4) a GDP adjuster (country per-capital GDP ÷ U.S. per-capita GDP).
• The volume weighted average net price across all countries.

The “limited option” would require, for each NDC-9 analog in each country, the weighted average net price; the total gross sales; the total net sales; the total sales units; the average net-to-gross ratio; the exchange rate; and the GDP adjuster.  Also required would be the volume weighted average net price across all countries.  One advantage of the limited option would be that it would not require sales and volume to be broken out by each price point in a country.

One glaring omission in the proposed rule and preamble is any direction on how manufacturers should calculate the various foreign net prices and weighted average net prices contained in their submissions.  This is in stark contrast to weighted average net prices required under Medicaid and Medicare – i.e., Medicaid average manufacturer price (AMP) and Medicare Part B average sales price (ASP) – for which CMS’s instructions consume scores of pages of regulations, preambles, and releases.  However, the preamble states that CMS will be issuing guidance on the various submission data elements and reasonable assumptions that may be necessary, among other things.  This guidance will hopefully provide direction on the calculation of foreign average net prices.  Even so, given the complexity and resources needed to calculate even one weighted average price (for example, AMP), the task of doing such calculations for a multitude of foreign prices will be daunting indeed.

Under the Default methodology, the International Benchmark would be the lowest country average price for the NDC-9 analog, and would not change for the entire GUARD Model period once it is set.  In contrast, under the Updated methodology, the International Benchmark would be the cross-country weighted average net price – not the lowest country price – and could change with manufacturer submissions in subsequent years.  As mentioned above, the greater of the Default Benchmark or the Updated Benchmark would become the International Benchmark.  CMS would then apply an adjuster of 102% if the Default method is used, or 104% if the Updated method is used (CMS explains that the greater adjustment could incentivize manufacturers to submit pricing data).

No conclusion can be drawn a priori on whether it will be advantageous for a manufacturer to submit international pricing data.  On one hand, relying on CMS to determine the International Benchmark under the Default method may be advantageous because CMS’s data sources will reflect list prices rather than the net prices submitted by manufacturers under the Updated method.  On the other hand, the limited option of the Updated method produces a cross-country weighted average net price for use as an International Benchmark, which will be greater than the lowest single country weighted average net price under the streamlined option, and may even be greater than the lowest country Default benchmark that CMS derives from published list price data.  The Updated method also has the advantage of the 104% adjuster.  A manufacturer’s decision whether to submit foreign pricing data will have to be guided by the international prices and discounting of the particular drug.

The GUARD Model rebate amount:  Once the International Benchmark for an NDC-9 unit is determined for a performance year, it would be compared with the performance year Medicare Net Price for the unit.  If the Medicare Net Price exceeded the International Benchmark, the excess would be the per-unit GUARD rebate.  The Medicare Net Price would be derived from the following operation:

• For each NDC-11 in the NDC-9 family, multiply WAC x units dispensed (from Prescription Drug Event (PDE) records), and add them together to obtain the NDC-9 total.
• Subtract aggregate manufacturer rebates paid to Part D plans for the NDC-9, taken from Direct and Indirect Remuneration (DIR) data in PDE records.
• Subtract aggregate Part D Manufacturer Discount Program rebates for the NDC-9.
• Divide by total quantity dispensed for all NDC-11s (from PDE records).

The Incremental GUARD Model rebate per unit:  A manufacturer who owed a GUARD Model rebate for a performance year might also owe a Part D inflation rebate for the same NDC9.  If so, and if the GUARD Model rebate were greater, the inflation rebate would be subtracted from the GUARD Model rebate.  Because an inflation rebate is owed for a fiscal year while a GUARD rebate would be for a calendar year, the inflation rebate would be converted to a calendar year using a time-weighting methodology.

Total Incremental GUARD Model Rebate for a performance year:  This would be the per-unit Incremental GUARD Model rebate multiplied by the number of units dispensed during the performance year, based on PDE data.  Units purchased under the 340B Drug Discount Program and units associated with compounded drugs would be excluded.  The Total Incremental GUARD Model Rebate would also be reduced for a drug currently in shortage, using a formula similar to the corresponding reduction under the Medicare Part D inflation rebate program.

Invoicing and payment process:  Manufacturers would be invoiced for the Total Incremental GUARD Model rebate separately from and at different intervals from the Part D inflation rebate invoices but using a similar process.  Final reports, which would also serve as invoices, would be sent 22 months after the end of a performance year, with two reconciliation reports at 12 months and 36 months, respectively, after the final report.  The reconciliations would adjust for updated claims and other reported data.  Payment of the initial rebate, and any additional amounts owed under the reconciliation reports, would be due within 30 calendar days after each report is received.  Thirty calendar days before each final report and reconciliation report, manufacturers would receive a preliminary report and have an opportunity to submit suggestions of error within 10 days following receipt.  Like suggestions of error under the inflation rebate program, GUARD suggestions of error would be limited to correction of mathematical errors, and the rule would preclude administrative and judicial review of CMS determinations of the Total Incremental GUARD Model Rebate, the dispensed units, or whether a drug is a GUARD Model Drug.  (Curiously, as authority for precluding review of the rebate amount or units, CMS cited the Part D inflation rebate statute, despite the obvious fact that this statute does not pertain to GUARD.)  In addition to the scheduled reconciliations, CMS could reconcile at any time after discovering its own error or learning of a manufacturer error, but could not do so later than five years after the date of receipt of the final report.

Enforcement:  Penalties for noncompliance would be the same as those under the Part D inflation rebate program.  Failure to pay an Incremental GUARD Model rebate within the 30-day deadline would subject a manufacturer to a civil monetary penalty (CMP) of 125% of the rebate (or reconciled amount) that the manufacturer failed to pay.  The penalty would be in addition to the underlying rebate owed.  The CMP notice, hearing, and appeal procedures would be the same as under the Part D inflation rebate program.  CMS notes that it could also refer non-compliance cases to the HHS Office of the Inspector General, the Department of Justice, or the Treasury Department.

Interaction with other government price reporting and discount programs:  With one exception noted below, GUARD Model rebates would not affect other reported government prices or government discounts in the U.S.  Drugs selected for Medicare MFP negotiation would be excluded from GUARD while the MFP is in effect.  Units purchased under the 340B program would be excluded.  Rebates paid under the Part D Manufacturer Discount Program would be excluded, as would voluntary rebates paid to Part D plans.  The exception is Medicaid rebate best price.  While Part D inflation rebates are excluded from best price by statute, the GUARD Model proposed rule does not explicitly exclude GUARD rebates from best price – nor could it do so, because CMS’s waiver authority under Social Security Act 1115A does not extend to waiver of Medicaid rebate best price provisions under CMMI demonstration models.

Effect on Part D enrollee out-of-pocket costs:  The GUARD Model is designed to reduce drug costs to Medicare, not necessarily to Medicare Part D enrollees.  CMS admits that the model will have only an indirect impact, if any, on enrollee cost sharing.  For example, it may reduce drug launch prices for GUARD drugs, which could have “cascading effects” of reducing co-insurance that is based on a percentage of list price, or causing Part D plans to change their benefit design to reduce enrollee cost-sharing.

Tomorrow’s post in FDA Law Blog will examine GLOBE, the Medicare Part B model for  MFN pricing.