JAMA: Withdraw This Flawed and Inaccurate Article About the 340B Program and Drug Prices

The Journal of the American Medical Association (JAMA) recently published Estimated Changes in Manufacturer and Health Care Organization Revenue Following List Price Reductions for Hepatitis C Treatments by Sean Dickson and Ian Reynolds. The paper purports to show that manufacturers’ net revenues increased following a decrease in three drugs’ list prices.

However, the paper contains a significant computational error: The authors do not properly calculate the 340B ceiling price. Consequently, the authors’ calculations are inaccurate and their conclusions are erroneous. As we might expect, a manufacturer’s revenue per-patient doesn’t increase when list prices decline and net prices remain constant.

Below, I explain how to compute a drug’s ceiling price under the 340B Drug Pricing Program. I then illustrate the crucial mathematical error in the JAMA paper. I also briefly highlight a few other errors and omissions in the paper.

The paper’s serious and irredeemable flaws imply that readers will be significantly misinformed about drug prices and the 340B program. You may or may not agree with me about the 340B program. But I hope you will concur that respected academic journals should publish truthful and accurate information.

#MATH

The authors state: “The 340B discount is 23.1% of a drug’s average manufacturer price.” This statement is not accurate. Therefore, the paper’s computations are unreliable and its conclusions are erroneous.

Understanding why Dickson and Reynolds are wrong requires a brief detour into the wonky underbelly of 340B pricing computations:

• Pharmaceutical manufacturers that participate in the Medicaid Drug Rebate Program (MDRP) agree to offer a 340B ceiling price to covered entities. The 340B ceiling price is statutorily defined as the Average Manufacturer Price (AMP) reduced by the Unit Rebate Amount (URA) from the MDRP. (source)

• The URA for brand-name drugs is defined as the greater of: (1) the difference between AMP and the best price, and (2) a statutorily-defined minimum rebate percentage equal to 23.1% of the AMP. (source)
  
  Manufacturers are also required to pay an additional rebate when the AMP increases by more than a specified inflation factor. For purposes of illustration, I ignore this additional component. Plus, list prices for some hepatitis C drugs have not increased, so this factor won’t affect the computations.

• Best price is statutorily defined to be “the lowest price available from the manufacturer during the rebate period to any wholesaler, retailer, provider, health maintenance organization, nonprofit entity, or governmental entity in the United States in any pricing structure (including capitated payments), in the same quarter for which the AMP is computed.” Like much else in government pricing, there are certain exceptions. (source)

With this background, we can revisit the author’s computations in Table 3 of their paper.

Consider the figures in Table 3 for Harvoni, which is shown to have had a WAC of $94,500 and a PBM rebate of $70,500.

• AMP is not equal to the wholesale acquisition cost (WAC) list price, but for the purposes of illustration, we will rely on the authors’ assumption that WAC approximates AMP. Note that AMP excludes commercial PBM rebates and certain other discounts, so it is generally much higher than best price.

• The authors compute Harvoni’s “340B Discount” (their terminology) to be $21,830 (= 23.1% * $94,500).

• However, the PBM rebate of $70,500 established a best price of $24,000 (=$94,500 minus $70,500). That’s because manufacturers typically include commercial rebates in best price. (See comments in the next section.) To compute the URA and get the 340B ceiling price, we must compare the following two figures: 1) $70,500 (= $94,500 minus $24,000) and 2) $21,830 (= 23.1% * $94,500).

• The URA was $70,500, since $70,500 > $21,380. The 340B ceiling price was $24,000, equal to a 74.6% discount from AMP (WAC).

By using the correct 340B ceiling price, the manufacturer’s per-patient revenue would be the same at 340B and non-340B facilities for all products shown in Table 3. The paper's results are therefore inaccurate. This crucial error undermines the entire paper and all of its conclusions.

A BEST PRICE DIGRESSION

In a private email to me, Mr. Dickson argued that commercial rebates to PBMs are excluded from the computation of best price in Medicaid. That’s also wrong.

Those who are not government pricing professionals can misinterpret the statutory language in 42 CFR § 447.505(c)(17) to mean that commercial PBM rebates are excluded from the computation of best price in Medicaid.

In practice, manufacturers almost always include rebates to PBMs on behalf of commercial plans and commercial lives in best price. Manufacturers include commercial rebates in best price because those rebates are “designed to adjust prices at the retail or provider level” (per the language of 42 CFR § 447.505(c)(17)). “Providers,” in turn, are defined to include health maintenance organizations, managed care organizations, and other entities that “provide coverage … to individuals for illnesses or injuries” (per 42 CFR § 447.505(a)).

Don't take my word for it. Instead, rely on John Shakow of King & Spalding. He's one of the country's leading legal experts on government drug pricing. Here's what he told me:

“In my experience, very few drug manufacturers do not include commercial PBM rebates in Best Price. They include them because it would be difficult to take the position that the rebates are not ‘designed to adjust prices’ to insurance plans.

If there is confusion about this issue, it’s because CMS in 2007 gave the impression that some PBM rebates might not be designed to adjust prices to insurers. That’s a hard argument to sustain in 2019, knowing what we know now about the relationships between PBMs and their clients.

Most drug manufacturers are risk averse when it comes to Best Price; without some kind of assurance (a one-on-one communication with CMS, for instance), almost all take commercial PBM rebates into account when determining Best Price.”

I also polled government pricing experts from around the industry. All of them confirmed that manufacturers include rebates in the computation of best price. Here are three representative statements:

• “Nearly all manufacturers include commercial managed care rebates in determination of Best Price based upon the industry’s typical interpretation of relevant regulatory guidance.” Jeffrey Baab, Executive Director, Professional Services, Integrichain

• “My experience is that many manufacturers include PBM price concessions in the calculation of BP because they believe that such rebates adjust prices to providers (e.g. HMO or MCO).” Miree Lee, Bio/Pharma Pricing, Contracts & Compliance Consultant, M. Lee Consulting, LLC

• “Rebates to PBMs based on commercial utilization are nearly universally included in Best Price by most manufacturers.” Jesse Mendelsohn, Vice President, Model N

OTHER ISSUES

I found other issues that JAMA should have required the authors to resolve prior to the paper’s publication. The paper:

• Misstates the pricing actions taken by Gilead Sciences. Gilead did not reduce the price of its hepatitis C products. In reality, Gilead started a wholly-owned subsidiary called Asegua Therapeutics to launch authorized generics of Harvoni and Epclusa. The list prices for the authorized generic versions are more than 60% lower than those of their brand counterparts. However, the list prices of Harvoni and Epclusa did not change.

• Incorrectly reports per-patient revenue for Zepatier. Merck recorded no net sales (after rebates) from Zepatier in the first quarter of 2018, prior to the product’s price cut. This means that net revenues per patient were $0, not the $16,543 shown in Table 3. This fact was widely reported. (See The New York Times coverage in Merck Is Lowering Drug Prices. There’s a Catch). There is no evidence to suggest that average net revenues in Medicare Part D differed from this overall average.

• Ignores how market competition affected the drugs’ list prices. The paper fails to consider the most likely alternative explanation for manufacturers’ pricing behavior. AbbVie launched Mavyret in 2017 with a wholesale acquisition cost (WAC) list price that was significantly lower than the list price of other therapies. Mavyret rapidly gained share and was the most prescribed product for the first half of 2018. See this market share chart.

• Fails to consider how list prices affect pricing and patients’ out-of-pocket costs. The authors provide a combined “Medicare and beneficiary Net Cost” computation. Putting aside the math errors outlined above, this computation ignores the significant effect of a lower list price on patient costs. For instance, a Part D beneficiary’s median out-of-pocket costs for Zepatier are $2,622 in 2019, compared with a median out-of-pocket costs of $6,338 for Harvoni. (source) The $3,716 difference in out-of-pocket costs is absorbed by the 340B covered entity. How is that sensible?

My bottom line: You will be greatly misinformed if you rely on this paper. Its publication shows a failure of the peer review process at JAMA. For the sake of the journal’s integrity, it should revise its peer review process and withdraw this paper from publication.