Smarter ways to boost drug innovation

We need to reduce costs

Joe Biden’s single best-polling (and most under-covered) public policy achievement is probably that, thanks to the Inflation Reduction Act, Medicare is currently negotiating bulk purchasing discounts for the 10 most widely used drugs that account for the largest share of Medicare expenditures. The pharmaceutical industry is currently suing in court, counting on Federalist Society judges to bail them out, and Republicans have, of course, promised to repeal the IRA if they win.

Meanwhile, this program — though significant — is limited, largely because of the insistence of outgoing senator Kyrsten Sinema. It strikes me as very unlikely that Democrats will control congress in 2025, but if that happens, it’s likely they will expand the program to cover more drugs as a deficit reduction measure.

The fact that this is a proposal to cut spending is a reminder that the politics of this issue are pretty odd. Liberals argue that the government is engaging in wasteful overspending by providing services to people at high unit costs rather than insisting on bulk discounts. Conservatives argue that paying the high prices is good because it creates a financial incentive for innovation, and that promoting innovation is more important than donning the green eyeshades. You could easily imagine transposing these arguments, with bargaining cast as the “neoliberal” position while paying the high prices is far-sighted “industrial policy” that emphasizes the importance of state-led growth rather than prioritizing strict efficiency or cost effectiveness.

The story of how we came to this particular alignment is interesting, and I’m not sure I fully understand it. But while this is obviously an important issue for politics and for budget policy, I wish we could get a more normal debate going about why pharmaceutical research is so costly and difficult in the first place. This is clearly a heavily regulated space, and there are good reasons for it to be regulated. But the particular ways in which it’s regulated go well beyond the clear public interest in ensuring the safety of medication to include all kinds of protectionism and slightly odd bioethics hangups. The process as a whole is not designed to pass any kind of cost-benefit test, and therefore is very costly. Everything about pricing and payments happens downstream of an excessively costly and time-consuming approval process, and the best solution for more innovation and lower prices is to address the root of the issue.

Why pharmaceutical pricing is weird

The basic issue with pharmaceutical pricing is that manufacturing and distributing pills is typically pretty cheap, but researching a new medicine and getting it through the approval process is expensive.

And drug development is more expensive than it looks ex post, because you don’t know in advance which projects are going to work out. The upshot is that any new medication you develop has high fixed costs associated with it, but low marginal costs of producing additional doses. Meanwhile, thanks to the patent system, you get to enjoy monopoly rights to sell your pill, which gives you a lot of pricing power. Most medications face at least some competition from other medications that do somewhat similar things. But still, if you’ve got a drug under patent and it’s good, then you generally want to charge much more than the marginal cost of manufacturing a pill for each dose — at least when you’re selling in rich countries where the ability to pay is high.

This creates a situation where there is a lot of upside to finding ways to force the price down. The simplest way is to just do price controls. The nation of Portugal can say “our expert panel of health professionals has determined that the social value of your medicine is such-and-such and that’s what you can charge.” Now, of course, the companies aren’t totally naive price takers. If Portugal tries to squeeze them too much, they can say “fuck off, we’re not going to sell the pill in your market.” So that becomes a bit of a regulatory bargaining dynamic.

But private health insurance plans can also bargain — and so can middlemen like pharmacy benefit managers.

Another idea, recently floated by Bill Ackman, is “to make it illegal for drug companies to sell the same drugs abroad for lower prices than they sell them for here.” So basically whichever country in the world gets the best deal on the price of a drug, the United States has to also get. Ackman’s idea is this will lead to “a globally negotiated price that will be lower than the prices that U.S. consumers pay now and higher than what foreigners pay now.” But this strikes me as more dangerous than the US just doing unilateral negotiation.

Imagine pricing strategy for HIV medications in this universe. Right now, a lot of medicine flows at very low cost to people in African countries, even as pharmaceutical companies charge higher prices in richer countries. If you ban price discrimination, you’d probably settle on something between the US and European countries, but companies would end up essentially exiting the market of serving poor people. If American wants to force prices down, it’s better for everyone that we do it explicitly and not through bans on price discrimination or through “reimportation” schemes where we buy Canadian medicine. Price discrimination is annoying, but it’s usually for the best when you have the disjoint between fixed costs and marginal costs.

I also have to say that I find the innovation-based argument against price negotiation to be a little underwhelming. Federal employees who travel for work need to follow specific guidelines on things like paying for a hotel room. But precisely because the federal government is such a large employer, they leverage the size of the market to secure a discount price for reimbursed travel from most hotel chains. Obviously it would be better for the hotel industry if the fed didn’t do that. But would it be better for society? It would lead to more investment in the hotel sector, but that’s neither here nor there to the extent that it just directs capital away from other uses.

If we want to see more pharmacological intervention, we should directly address the barriers to bringing new medicine to market.

Cost-benefit for drug approvals

In “Contagion,” the pandemic ends when a heroic scientist injects herself with an experimental vaccine, deliberately exposes herself to the virus, and then doesn’t get sick.

It’s a good movie, but anyone with a passing familiarity with statistics or the scientific method could tell you that an experiment with a sample size of one is not actually very informative. A person treated with a totally useless vaccine might avoid infection just through sheer dumb luck. And a reasonably effective vaccine still won’t be flawless. You can’t really understand side effects and downsides without a reasonably large sample. It would be boring and un-cinematic to walk through all the statistical ins-and-outs, but in the real world, the larger your sample, the better your information. At the same time, the larger your sample, the slower and more expensive your trial. You probably don’t want to approve new drugs based on a sample size of one. But you could have a sample of 10,000 people and still some unanswered questions about rare side effects. At a certain point, you need to decide your information is good enough to run with it.

But what point is that?

The FDA currently errs very strongly on the side of a precautionary principle. As a question of bureaucratic logic, this is pretty easy to understand: If you approve a drug quickly and then it turns out to have bad side effects, that’s going to be a big embarrassment for the FDA. But i you drag things out for an extra two years, asking for more data, you’ll have a frustrated pharmaceutical company, but no big public backlash. I used to hear a lot from right-of-center people that this was a bad approach and the FDA was too risk-averse, but we then saw with the Covid vaccines that right-wingers themselves can get quite paranoid about speedier drug approvals.

Still, the fact remains that the clearest way to boost pharmaceutical innovation is to take a more balanced approach to decisions under uncertainty.

With anything like this, you can make Type I errors where you are too permissive and allow drugs onto the market that have serious downsides, or you can make Type II errors where you are too restrictive and block a drug that has significant benefits. Common sense says that you should worry a lot about Type I errors when the condition being treating is relatively trivial (you don’t want people getting heart attacks because they took a pill for baldness), but worry less when you’re talking about a very serious illness. If you’re at very high risk of dying in the very near future from a fatal disease, then you probably aren’t as worried about the downsides of experimental treatments.

The current FDA process does not formally incorporate this kind of consideration into their decision-making. In a 2017 paper, Leah Isakov, Andrew Lo, and Vahid Montazerhodjat did a bunch of math (“Bayesian Decision Analysis”) based on the official Burden of Disease data. They showed that regulators are much too risk-averse when it comes to approving new treatments for terminal ailments that currently lack new treatments. In other words, you very rarely see a new drug rushed onto the market only to be withdrawn after downsides emerge. From the standpoint of FDA public relations, that’s a win. But it’s like watching a target shooter who never shoots below the bullseye. Obviously I get why it would sound odd, politically, to be the guy who advocates for more harmful side effects. But there are real benefits to bringing new medicines to market — that’s why people worry so much about the cost of prescription drugs — and lowering the barriers to approval is one of the best ways to make that happen.

Going global

Science is a global enterprise, and obviously most of the medicines used in the United States are also used in foreign countries. By the same token, lots of clinical trials happen in foreign countries and data from foreign clinical trials is routinely submitted at part of drug approval processes.

I’m told, though, that the FDA generally frowns on the idea of making a submission based entirely on foreign data.

This isn’t a rule per se, more just a thing people know about. You can see some of that in this explainer from Windrose Consulting Group, which talks about a particular submission based on Chinese data that the FDA rejected. Now, part of the rejection seems to be skepticism about the reliability of clinical data from China. And fair enough; obviously you should not approve drugs on the basis of unreliable data. But there’s also a somewhat fuzzy FDA stipulation that foreign data is acceptable only if it is “applicable to the U.S. population and U.S. medical practice,” which according to Windrose encompasses “questions around how genetics, lifestyles, environments (living, working, etc.), and the homogeneity of populations impact how individuals, let alone entire subpopulations, react to treatment.”

This is all, on some level, reasonable.

But we would benefit from actually doing the work to resolve these concerns, rather than leaving them hanging out there as hazy sources of uncertainty. This report from IQVIA tries to offer better advice about navigating the currently murky landscape and makes it clear that plenty of foreign data is currently being used. Still, clinical trials are in the aggregate pretty US-centric, even though they are labor intensive, which means that doing them in foreign countries where incomes are lower is probably going to be cheaper. And to be clear, “countries where incomes are lower” means basically every country other than Norway and Switzerland — it’s not like the only options to save money is to look to the poorest populations on earth. We should work to deliberately cultivate regulatory relationships with foreign countries so that reliable trials can be routinely conducted in Bulgaria or Jamaica or Honduras. You of course want to be aware of genetic differences between national populations. But the US is itself a very diverse country, and we currently have problems recruiting adequately diverse populations into our clinical trials. Going abroad could be a good solution to that.

Fostering innovation

What I’m struck by is the current disjoint in American approaches.

When it comes to the pricing of drugs, we treat pharmaceutical company profits as sacrosanct because encouraging pharmaceutical innovation is so important. But when it comes to the regulation of new drugs, we treat the upside of new medication as trivial. Clearly it’s true that unscrupulous pharmaceutical companies could try to smuggle unreliable foreign clinical data under the FDA’s noses and that strong financial incentives exist to bring to market drugs that may be unsafe for certain populations. This is a regulated area for good reason.

But treating regulatory delays as costless, then paying the piper for massive windfall profits because we want to encourage innovation doesn’t make sense.

We should be working proactively to create pipelines of reliable clinical data from foreign countries where trials are cheaper to perform, and we should be taking a more balanced approach to the costs and benefits of requiring more data. Reducing the cost basis for developing new drugs will encourage innovation, mechanically pull prices down via competition, and let us worry less about the downsides of driving a hard bargain on behalf of the American taxpayer.