So I opened my email today and found an email from Joseph DiMasi, the Director of Economic Analysis at the Tufts Center for the Study of Drug Development. He is the researcher who did the original work I referred to yesterday in determining the cost of drugs.
I can’t tell you how happy that makes me. Too rarely people who do the actual research get to be the ones to talk about it. As I (and others) have said, it’s important for the people who do the work to be able to defend it. It’s also good because it’s only by hearing opposing arguments that I (and you) learn.
Here’s the truth about research: none of it is perfect. All research has limitations, even the best. And we have to judge how much those limitations can bias or affect the outcomes of our research. The importance of peer-reviewed research isn’t just that it’s been graded adequate by others; that has its own flaws. Even more critical is the importance of transparent methods so that we all can see how the study was done.
Yesterday, I channeled Dr. Angell and Public Citizen to critique Dr. DiMasi’s work. Dr. DiMasi today directed me to some published defense of his work. I offered him the chance to respond directly, but he is going to let me take a crack at it first and then respond again if he wants to.
Understand: this is my interpretation of his work. You should go read it for yourself. Assessing Claims About the Cost of New Drug Development: A Critique of the Public Citizen and TB Alliance Reports.
Since I stuck to the Public Citizen critique, that’s what we’ll discuss here. Here are the top points from the Executive Summary.
1) Pharmaceutical R&D is an investment with expenditures made years before any potential returns are earned. Based on standard principles in economics and finance, these investments have opportunity costs that are real and highly relevant. The time costs associated with new drug development are inappropriately ignored in their entirety in the Public Citizen report.
Dr. DiMasi maintains that the opportunity costs are improperly removed by Public Citizen. We’re not going to resolve this one. I understand Dr. DiMasi’s argument, which is that since there is delayed time until the drug yields any reward it is inherently less valuable than an investment with immediate returns. Perhaps my auto industry quip was to cute. I still maintain that since research is not a choice – they have to do it – this is not an especially compelling argument. However, people do come down on both sides of this. I think I side with Public Citizen more on this point. Even if that’s how economists do it in ledgers, it’s not convincing to me that we should double the actual cost of investment because pharmaceutical companies decided to research drugs, and it takes a long time.
2) The Public Citizen report, noting that R&D expenditures are deductible under the corporate income tax, maintains that R&D cost estimates should be reduced in percentage terms according to the corporate income tax rate. The estimates in our studies were meant to examine trends in private sector resource costs, and changing tax structures mean that after- tax costs can mask such trends. The Public Citizen perspective, however, also reflects a fundamental misunderstanding of the nature of the corporate income tax. Profits (i.e., net income) are the target of the tax, not gross income. Deducting business costs is just the mechanism by which the targeted tax base (profits) is determined.
Dr. DiMasi claims that public citizen, by adding 35% to all costs, simplifies the pre/post tax argument. I think that’s fair. It is likely that not all of the costs, if the were not research, would have gone to profits. However, given the vast amounts they already spend on marketing, it’s likely that some of those costs would have become profits and not gross income. Overall, though, this feels like Public Citizen is using an argument not totally dissimilar to what Dr. DiMasi was in his first point. They are assuming pharmaceutical companies have a choice to do research or make money. If we say they must do research, then we can’t say they’re getting a tax break for choosing to do so. I side with Dr. DiMasi on this one.
3) Public Citizen used published annual data on industry R&D expenditures from the industry’s U.S. trade association and FDA data on the number of new drug application (NDA) approvals to measure pre-tax out-of-pocket R&D costs. However, they used incomplete and mismatched data to derive the ratios that resulted in their cost estimates. The numerators of their ratios exclude much relevant expenditure and the denominators are inflated by including approvals of firms that did not contribute expenditure data to their numerators. For these reasons, their estimates using published data are deeply flawed and substantially understate R&D costs.
This is actually a valid point by Dr. DiMasi, but I’m not sure it changes things as much as he says. He is correctly pointing out that some of the drugs approved by the FDA were not made by PHRMA companies; not all companies belong to the trade association. Therefore, since Public Citizen used PHRMA R&D numbers, the amount invested is underestimated for the number of drugs approved by the FDA. However, most big companies making the major investments do belong to PHRMA. He claims that 29% of drugs approved from 1994 to 2000 were non-PHRMA companies. I’d need to see compelling data for 2000-2003, which I presented. And even then, I’d want to see how much was spent on research by the non-PHRMA companies. However, I concede the point. Public Citizen’s numbers are biased towards a lower cost-per-drug.
4) Public Citizen also used the NDA as its unit of observation, as opposed to a new drug (i.e., a new active ingredient). This is both technically and conceptually inappropriate. Many of the NDA approvals are not for new molecular entities (NMEs). However, many of these approvals are also not for new product presentations and/or are obtained by firms that have no relationship with respect to the drug in question to the sponsor of the original NME approval. On a conceptual level, the costs of obtaining non-NME NDA approvals on line extensions are intimately related to the costs of the associated NME NDA approvals. The most appropriate perspective to take on the R&D process is to use a new drug (active ingredient) as the unit of observation and examine costs over the lifecycle of the drug.
Dr. DiMasi is arguing (correctly) that not all new approvals by the FDA are for new active ingredients. I could not agree more. As I previously discussed, too much of the time they are for me-too drugs. Sometimes, they are for new indications of drugs, which require less research comparatively. Dr. DiMasi argues that we shouldn’t count these drugs in the calculation since they aren’t really new drugs and lower the apparent cost-per-drug.
I’d be more sympathetic to this argument if the commercials said, “it costs $800 million to develop a completely new drug that does something no drug before it has done.” But they don’t. They say “it costs $800 million to develop a drug.” Since they want the public to believe that every slight change to an old drug is a new and important drug that you “must discuss with your doctor”, then those drugs count too. Some new indications for drugs even result in new names for drugs and different colored pills. The companies pass these off as new drugs, and report sales on them differently, so I’m not totally disinclined to count them as separate in the denominator.
So where does this leave us? I have to make a decision as to how to judge these different arguments. Since I think most people think about the cost of R&D as how much money companies have to shell out to make drugs, I’m not persuaded that opportunity costs should be fully counted. I’m swayed, however, that adding 35% to it for avoided income tax is not warranted. I believe that Public Citizen did underestimate the cost by including drugs from non-PHRMA members. I’m not entirely convinced that we should ignore new formulations or new indications, though. You can reach your own conclusion.
So Dr. DiMasi’s study said $802 million. Public Citizen said $110 million. I bet the real number is somewhere in between.
The truth of the matter is that this ignores the larger argument. I have no problem believing that pharmaceutical companies spend a lot of money when they invest in novel research to lead to NME that do new things. But that’s not the majority of their business. Nor is it a majority expense. So talking about only those types of drugs and talking about only that expense ignores the real issues. If – tomorrow – pharmaceutical companies announced that they would no longer focus on me-too drugs and would invest massive money into the development of NME only, I’d be the first in line to defend them and argue for big breaks for them. But they aren’t.
Moreover, if they want people to know the true average cost for development of a new drug, they could open their books, allow researchers to randomly select a number of drugs from them, and be transparent about the whole process. They aren’t going to do that. If they won’t, they can’t be surprised when people are skeptical of their claims.
P.S. Evidently I said that Dr. DiMasi had a medical degree. We can argue whether that is a compliment or an insult, but I was mistaken. He has a PhD. My apologies!