Skip to Content

Clinical Trials

That $43 Million R&D Figure

One of the readers in the comments section to the last post noticed Rebecca Warburton trying to clarify that absurd $43-million-per-drug R&D figure. You’ll find her response in the comments section to the Slate piece that brought this whole study so much attention. Says Warburton:

. . .Our estimate of $59 million is the median development (the “D” in R&D) cost per average drug, not just NMEs (new chemicals) and does not include basic research costs, for which there is no reasonable estimate available.

But that explanation won’t wash, as some of the readers over at Slate noticed as well. If you read the Light and Warburton article itself, you find the authors talking about nothing but “R&D” all the way through. In the one section where they do start to make a distinction, they brush aside expenses for basic research, on the grounds that drug companies hardly do any:

Companies under pressure from quarterly reports have difficulty justifying long searches for breakthrough drugs to investors. . .Little company R&D is devoted to basic research. Although industry association reports, based on unverified numbers from its members, claim that companies invest on average 17–19 per cent of sales in R&D, the most authoritative data come from the long-standing survey by the US National Science Foundation (2003). Its data document that pharmaceutical firms invest 12.4 per cent of gross domestic sales on R&D. Of this, 18 per cent, or 2.4 per cent of sales, went to basic research. More detailed reports from the industry indicate the percentage of R&D going to basic research is even smaller, about 9.3 per cent (or 1.2 per cent of sales) (Light, 2006). Thus the net corporate investment in research to discover important new drugs is about 1.2 per cent of sales, not 17–19 per cent.

So no, claiming that the $43 million figure is only supposed to represent the “D” part of R&D is disingenuous. There’s another line from this paper, quoting Marcia Angell, that I think gets to one of the roots of the problem with the way these authors have characterized drug research. Angell is quoted here with approval – everything she and Merril Goozner have to say is quoted with approval:

It is also unclear how far back one should go to count up the costs of discovery, given that often there are several strands of research that are pieced together. In Angell’s view, the critical step in ‘discovering’ a new drug is understanding how the disease works and finding one or two good targets of vulnerability in the defences of a disease for intervention. Basic research ‘is almost always carried out at universities or government research labs, either in this country or abroad’ (Angell, 2004, p. 23).

And there you have it. The critical step is understanding how the disease works, you see, and finding one or two good targets. By that definition, the vast amount of money that gets spent in the drug industry is then non-critical. This is a viewpoint that can only be held by someone who has never tried to discover a drug, or never held a serious conversation with anyone who has.
Let’s poke a few holes in that worldview. First off, if we waited to “understand” diseases before trying to develop drugs for them, we’d hardly have a damned thing on the drugstore shelves. Look at Alzheimer’s – the medical community is still having fist-waving arguments about its cause, while drug companies continue to sink piles of money into trying to treat it. (Almost all of which has gone down the tubes, I might add, and I helped flush some of it through myself, earlier in my career).
Then you have to find one or two good targets. Peachy! Where do you find those thingies, anyway? And how do you know that they’re good targets? I wish that Marcia Angell, Donald Light, or Rebecca Warburton would let the rest of us in on those secrets. As it is, we have to take chances on some pretty tenuous stuff, and often the only way to find out if a target really has any connection to human health is to. . .well, to discover a drug candidate that hits it. And develop it, and get it through tox, and into humans, and through Phase I, and into Phase II, and more likely than not these days, into Phase III before you really find out if, you know, it was actually a good target. We pass on those results to the rest of the world at that point. But that doesn’t count as research, apparently.
And how about the drugs that have been developed without good mechanisms or targets at all? Metformin, ezetimibe, rosiglitazone and pioglitazone: none of these had any detailed mechanisms worked out for them while the money was being spent to develop them. These are the sorts of things we do around here in between having meetings to decide what color the package should be, and right after we do that thing where we all jump around in rooms knee-deep in hundred-dollar bills. Exhausting stuff, that money-wading.
But what I’d really like to ask Light and Warburton about is this: if you do think that the Tufts/diMasi estimate is crap, why did you feel as if the antidote was more crap from the opposite direction? Honestly, I’d think that intelligent people of good will might be more interested in decreasing the total amount of crap out there instead. . .

17 comments on “That $43 Million R&D Figure”

  1. luis says:

    If it’s so cheap and easy, I’m sure that in 10 years, the NIH translational institute will have a strong track record of getting new drugs out. They don’t have the same distorting pressures that drug companies have.
    On the other hand, they might realize that it’s not that easy and you will have been proven right (at the cost of a few billion for the taxpayer).

  2. MoMo says:

    Talk is Cheap and that’s all we get from the business bunkers and their chattering monkeys. The scientists in the trench that have been through the process should write these papers on the actual costs, but have any of these academics considered this? Nooooo!
    They forgot all sorts of variables, but then, they have to justify their existences somehow.
    Why are we even debating this?
    Drug discovery is all situational-specific.

  3. Hap says:

    It’s worth debating because 1) their output will be used to determine policy (probably – even if the Dems plan gets sunk by the Supreme Court, there will probably be something else that will impinge on drug development) and so refuting bad and/or dishonest numbers will help (maybe) that policy is made based on reasonable data and 2) financial people need to have an idea how much drug development costs to invest in it. Even if drug development costs are situational (the variability of which is still helpful to know), an order of magnitude estimate paired with a probability of making it to market (and an idea of how much it could make if it gets there) is necessary for someone to put their money up. No money, particularly in a world where big pharma is turning over research to VC-dependent little pharma, means no new drugs, eventually.
    The probablem with political numbers is that the people touting them don’t seem to care if they’re right or wrong, as if their correctness were determined by the hardness and identity of their political opinions and not by the things they allegedly represent.

  4. oleginger says:

    @3 – Hap
    I think it does not really serve the discussion to simply categorize the opposite opinion as motivated by particular party politics. The arguments and data presented in the paper are so weak it would seem perfectly sufficient to argue on the basis of better data and deeper understanding of the drug discovery process.
    When it comes to sticking numbers to the various phases of R&D it’s also interesting to get back at this paper by Paul et al ( Their model is better documented than the Tuft Center’s and calculates even higher costs. On the other hand it allows for a more detailed discussion, too.

  5. Hap says:

    No, but if you submit numbers that don’t pass a basic smell test, then something is driving your numbers, and without a lot of data, it’s probably not reality. I would probably generally agree with their politics, but I would prefer they be based on reasonable and accurate numbers rather than some sort of fantasy.
    If their point is that the cost numbers are arbitrary and politically determined in general, then that would seem to be a legitimate and problematic point for everyone involved, but the response seems to indicate that that wasn’t their point.

  6. Glen says:

    The critical step is “understanding how the disease works?” I had no idea we knew so much about disease mechanisms. It has been my understanding that we may have a hypothesis regarding an underlying mechanism, but that our overall understanding of any disease is always changing.

  7. Luigi says:

    Take Angell, Light, and Warburton, take a pinch of Francis Collins and you have a a quartet of no nothings about drug R & D and a total recipe for disaster.

  8. DrugA says:

    It has always seemed to me that a more robust approach to determining the cost of drug development would be to analyze small companies developing their first product to the point that the drug is either approved or out-licensed (or the company acquired). That way, you would have a very real estimate: capital raised to achieve one product (in the case of out-licensing or acquisition, you’d have to add on the investment to take it the rest of the way through the approvals process, but this should be a good deal more measurable). What this wouldn’t give you is the cost of the drugs that don’t succeed, nor would you have the inefficiencies associated with a large pharma bureaucracy. You could deal with that by taking the number as a minimum estimate. An analysis of small companies might bias you towards innovation that originated in academic labs, but, as Derek points out, that’s just the start of the journey.

  9. M says:

    Its Wall Street versus Big Government!
    Sure its comedy at its finest, but Joe Peasant has bigger problems: can’t afford gas in his tank.
    Keeping Granny alive to 100 only makes the problem worse.

  10. Cytirps says:

    Hopefully, the $43MM or $59MM they were talking about is just referring to chemically development. Hopefully, they are intelligent to know that clinical development is the big spender. Nowadays, clinical trial of any cardiovascular drug would require at least hundreds of million. It makes me shiver just to imagine recruiting 30,000 patients for the clinical trial.

  11. CMCguy says:

    Thanks Derek as I think your last lines well captures the basic problem concerning the Tufts calculations being debatable they choose to use extremist positions to counterbalance and support their predetermined conclusions (Given away by citing positively Angell and Goozner “source facts”). These are dealing with economics models so seems Light/Warburten decided to go against convention as to apply crap in to their system in expectation of getting something else out.
    I could almost believe a $43M estimate for “R” but to attribute only to “D” is incredulous because of the costs of clinical trials where real money involved.

  12. Licensed says:

    @DrugA: A very good point that was also made in the previous post. When companies pay $300M for Calistoga, or Plexxikon, or license Dimebon for $500M or Metabolex at $100M or Exelixis or Array for $100M, that is providing Market driven data on what companies think these assets are worth. The people signing these checks have negotiated the deal for what it is worth to them. A Ph II asset that has been significantly de-risked is still worth $300M even if you still have to run Ph III trials that may fail.

  13. Rick wobbe says:

    I’ve read the Light and Warburton paper and followed this blog and the comments on Slate as well, including Warburton’s and Light’s contributions. I’ve also read all the DiMasi et al papers. But I’m STILL waiting for an unbiased analysis!
    The choice of words and references in the Light and Warburton paper is poor enough (i.e. unprofessionaly inflammatory and woefully incomplete, respectively), that it naturally calls into question the motivation of the authors, the quality of their work AND the standards of the journal that accepted the manuscript as (apparently) an article rather than an opinion piece. Any valid points the piece has(and I believe there are some) are pre-emptively undermined by the decidedly oppositional tone adopted from the very outset of the article (i.e. the wording of the title).
    There are a number of other academic economists who have raised, in mercifully UNheated and objective scholarly articles, serious and legitimate criticism of the TCSDD methodologies, especially the way they use cost of capital and their choice of corrections for inflation. In particular, Boston University School of Management’s Iain Cockburn and Harvard Business School’s Gary Pisano have written articles that warrant legitimate and appropriately skeptical scrutiny of PhRMA/TCSDD publications (and, in my opinion, the Light and Warburton paper also, since it appears to be based on the TCSDD findings).
    Altogether these studies convince me of one thing: we have not yet seen a well-reasoned, unbiased, economically sound figure for the mean or median cost to develop any drug (new, me-too, reformulated, recombined or even placebo). What I’m seeing now is a lot of heat but no light (no pun intended). My opinion, based on the sum total of what I’ve read from a wide range of sources, is that the most useful number lies (at most) roughly midway between the TCSDD and Light/Warburton extremes.

  14. Rick says:

    @ 4 – oleginger
    Important note about the Paul et al Nature Reviews Drug Discovery article you cited:
    Author affiliations
    1.Lilly Research Laboratories, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, Indiana 46285, USA.
    I’m just sayin’

  15. AKS says:

    I think others have noted that to get a basic idea of how much a drug costs you could just divide total R&D costs by number of new drugs achieved for that R&D spend. If you look at the figure in the recent NY Times article:
    They give the total number of new drugs approved per year going back to 1995, and also the total R&D spend per year going back to 1995. Even just eyeballing these figures, about $1 billion per approved drug looks about right (actually a bit on the low side; and its getting even more and more expensive more recently, with R&D spending going up but number of new approved drugs going down significantly from the seeming “glory days” of 1995 to 2000, although biologic drugs aren’t included here and could explain some of this). Maybe this is a rough calculation, but it can’t be too far off could it?

  16. davem says:

    I’ve read this article and the previous with interest.
    I have a medical background and would generally agree with the comments of other that the figure is going to be well short of the mark.
    However if you took a single innovative discovery made by a student, who then created a company to exploit it… maybe, just maybe $43m would be about right.
    Of course you aren’t considering any of the other students who ‘failed to make that amazing discovery’ but went on to show certain targets where pie in the sky.
    I always think of the line in the awakening film where Robin Williams (playing the part of sachs) is questioned about his research into
    ‘trying to extract the myelin protein from [some microorganism’
    to which his is response was
    ‘Yes, I failed, but at least I proved it wasn’t possible’
    I suspect the $43m figure forgets the studies where ‘it was proved not possible’, or the drugs that ‘aren’t as effective as the current best drug, and only marginally more effective than the placebo’ and then simply sidelines these as ‘not part of R&D.
    A more acurate way to measure the cost of a single drug maybe to consider the following.
    Profit made from ALL pharmaceutical companies (including those that only make generic drugs that are now out of patent) = X
    Investement into drug research (from governments, not for profits, pharma etc etc) = Y
    The total number of drugs on the market (old and new) = T
    Now the cost of devopment of drug D = (X – Y) / T
    Obviously this needs to be taken in advisement, as a lot of those drugs are going to include the paracetamol and asprin tablets that are sitting in my medicine cabinet.
    All I can say is that $43m sound way too low, as the original figure of $800m sounds far too high.
    We will probably never know the reality…

Comments are closed.