A big limitation to an approach to drug pricing used by many ex-US countries

While our attention is on ex-US drug prices, let’s discuss a huge limitation to an approach that many ex-US countries use in their pricing and access decisions.

TURN THE PAGE

Many ex-US countries explicitly or implicitly use cost-effectiveness analysis when making decisions about a drug’s price and whether or not there will be government access to it. Cost-effectiveness analyses have taken heat for their use of the quality-adjusted life year – a measure deemed discriminatory for assigning a lower quality of life in life extension to the elderly, disabled, and terminally sick as compared to someone who is younger, not disabled, or not sick. As such, the use of “evidence based on quality-adjusted life years (QALYs) in the process of negotiating a maximum fair price is not permitted” for the Medicare drug price negotiation program. This is important to remember when considering the use of ex-US drug prices in the US, but it is not the topic of this newsletter.

This newsletter is about another limitation of cost-effectiveness analysis and that is the misrepresentation of total added costs. The overwhelming majority of cost-effectiveness analyses conducted to inform the value of a prescription drug ignore the potential for a drug’s price to change—for both the intervention and comparator. Rather it is common practice to hold a drug’s price static over the entire model time horizon which often extends many years into the future (even past the expected exclusivity period when generic or biosimilar competition is expected to enter and drive down pricing). Ignoring these expected price changes can overestimate or underestimate the added costs of the new drug. This may not seem like a huge deal, until you consider that decisions to pay for and provide access to drugs are informed by these analyses.

In a newly published Value in Health article I authored with Drs. Josh Cohen, Peter Neumann, Tyler Wagner, and Jon Campbell, we show that incorporating future price changes across the product’s lifecycle can produce substantially different cost-effectiveness estimates. These differences are more pronounced for chronically administered treatments as compared to one-time treatments.  

For a chronically administered treatment, the cost-effectiveness estimate that assumed drug prices would remain static (as is common practice) resulted in a 60-80% less favorable cost-effectiveness estimate than the analysis that incorporated future price changes with stacked cohorts.

For a one-time treatment, the cost-effectiveness estimate that assumed drug prices would remain static resulted in a 25-35% less favorable cost-effectiveness estimate than the analysis that incorporated future price changes with stacked cohorts.

These substantial differences between the cost-effectiveness estimates that do and do not account for the expected changes to a drug’s price illustrate the importance of addressing this limitation of conventional cost-effectiveness analysis.  

We also identified which dynamic model input had the most influence on the cost-effectiveness estimate by assigning a wide range to each dynamic input. For chronically administered treatments, the dynamic input with the greatest influence on the cost-effectiveness estimate was the post-loss of exclusivity price change. For one-time treatments, the age of individuals at baseline and the discount rate had a greater impact on the cost-effectiveness estimate than the drug’s assumed price changes.

These differential impacts between chronically administered treatments and one-time treatments suggest that cost-effectiveness analyses that do not account for exclusivity periods and dynamic inputs favor certain types of treatments.

As we suggest in the publication, “by incorporating these price dynamics within CEA, we take one step toward modeling the competitive marketplace that, in turn, can help level this playing field and promote dynamic efficiency. We acknowledge the uncertainty that comes with making assumptions about price changes in the future and yet believe that the same sensitivity analyses that capture uncertainty for conventional inputs can be used for dynamic inputs.”

SO WHAT

If you ask any biopharma investor or innovator, none of them would tell you that they expect a drug’s price to remain static forever. The assumption of static pricing over time ignores the market realities of branded and generic/biosimilar competition. Omitting expected price changes in a cost-effectiveness analysis risks miscalculating the incremental treatment costs and misinforming coverage and reimbursement decisions. I understand that all models are wrong, but if we want them to be useful, they should at least attempt to approximate reality.
Below I present, and challenge, three arguments that are frequently cited as reasons against incorporating price dynamics in cost-effectiveness analyses.

Argument #1: Drug price increases following a drug’s launch will cancel out drug price declines after the loss of exclusivity. When cost-effectiveness analyses hold a drug’s launch price static forever, they may not account for price changes after the exclusivity period, but they also are not accounting for potential price changes over the exclusivity period. Drug prices can increase while the manufacturer has a monopoly over the drug and thus the increases over the exclusivity period could cancel out the decreases after the exclusivity period.

Challenge to Argument #1: Evidence refutes this argument. First there is the recent peer-reviewed analysis by Ching-Hsuan Lin and colleagues that shows “the average inflation-adjusted mean annual drug price change was -4.7%” over the exclusivity period. The whole paper is worth a read, but they analyzed 32 branded large-market drugs, and on average the net prices were declining over the exclusivity period. This challenges argument #1 if there are no price increases over the exclusivity period. A price decrease over the exclusivity period is not going to net out a price decrease after the exclusivity period.

Now in the paper by Ching-Hsuan Lin and colleagues, there were some drug characteristics that did suggest a price increase over the period of exclusivity period. But were the price increases large enough to offset the substantial price decreases that should occur after the exclusivity period? It’s doubtful.

The purpose of our CPE Exculsive on Tecfidera was to explore this very thing. Tecfidera’s list price nearly doubled over its exclusivity period. However, within six months of generic competition, generics were priced at less than 90% of the list price. The decrease after exclusivity was far greater and more immediate than the gradual increase over the exclusivity period, further challenging the argument that the two will cancel each other out.
One other counterpoint to this argument that we raised in a Health Affairs Forefront article was that the Inflation Reduction Act (IRA) “limits price increases following a drug’s launch by requiring rebates to Medicare if a drug’s price increases faster than inflation.” Thus, this limit on price increases following a drug’s launch also challenges argument #1.

[To be transparent, argument #1 is the argument I used to make when I held drug pricing static in the models I created. However, recent evidence, policy, and engagement with stakeholders evolved my thinking.]

Argument #2: There is great uncertainty around future drug pricing and no guarantee the price will drop after exclusivity.

Challenge to Argument #2: By holding a drug’s price static (and not varying this assumption), the model is assigning certainty to no future price changes which is not reflective of reality. With certainty, we know that patents will expire and exclusivity periods will end.
I acknowledge the variability around specific inputs related to future drug pricing (e.g., length of exclusivity period, price changes over exclusivity period, prices changes after exclusivity period), but economic models are designed to account for variability in model inputs. Economic models readily account for uncertainty and variability in clinical model inputs, but this practice has not extended to treatment costs for some reason. Exploring the impact of variability in model inputs on the cost-effectiveness finding is a core component of economic models.   

Our Value in Health publication described in the section above set out to do this very thing for model inputs specific to future price changes. As can be seen in Figures 1 and 2 in our article, a wide range of potential values for each dynamic cost input can be assigned and variability can be quantified in sensitivity analyses.

One other counterpoint to this argument that we explored in a Health Affairs Forefront article was that “the IRA mandates “negotiated” price reductions for certain drugs even before drugs lose exclusivity, which also challenges the argument that future price reductions are too uncertain to incorporate into CEA”.

Argument #3: Accounting for genericization in a cost-effectiveness analysis allows companies to take credit for genericization to recapture savings intended for society.
Challenge to Argument #3:  By omitting genericization, we are creating an unlevel playing field between interventions and an unlevel “reward” to companies while also misrepresenting the incremental treatment costs. The “reward” that the companies can take credit for is dependent on the time on treatment over the exclusivity period. Without accounting for the exclusivity period in economic models, we aren’t capturing this dependency.

The spirit of this argument is grounded in dynamic efficiency which is a goal I support. We want to use resources efficiently and get the optimal amount of innovation. However, the evidence does not exist today to know the optimal amount of the “value” that should be allocated to the company versus the amount of the “value” that should be allocated to society. This is not specific to the incorporation of genericization into an economic model. This is an issue with any cost-effectiveness analysis, even those that assume static costs.

[To be transparent, I believe argument #3 is pointing out an important point about value allocation between the innovator and society. However, I believe it is falsely made against incorporating dynamic costs into cost-effectiveness analysis, when it is applicable to any type of cost-effectiveness analysis used to inform the optimal reward for the innovator. Unfortunately, we don’t yet have robust and accepted evidence on the optimal share of value for the innovator and optimal share of value for society. This is an important area of future research for the use of cost-effectiveness analyses to inform resource allocation decisions.]
One final thing to bring this back to international price comparisons. The headlines are full of stats that the US pays 3 times that of other high-income countries for prescription drugs. Importantly, that stat is for branded drugs. I haven’t seen much coverage that prices in the US are “lower for unbranded generic drugs:  Americans paid $0.67 for every dollar paid in comparison countries” and that “generics represent 90% of prescriptions filled”. I understand why this has not been an emphasis given the sentiment of the executive order around global freeriding, but just as cost-effectiveness analyses need to do a better job acknowledging that a drug’s price changes dramatically over its time in the market, so do we.