About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases. To contact Derek email him directly: Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
Not Voodoo

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
Realizations in Biostatistics
ChemSpider Blog
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Eye on FDA
Chemical Forums
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa

Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
Gene Expression (I)
Gene Expression (II)
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net

Medical Blogs
DB's Medical Rants
Science-Based Medicine
Respectful Insolence
Diabetes Mine

Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem

Politics / Current Events
Virginia Postrel
Belmont Club
Mickey Kaus

Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« The Infinitely Active Impurity | Main | Bexarotene for Alzheimer's »

February 10, 2012

The Terrifying Cost of a New Drug

Email This Entry

Posted by Derek

Matthew Herper at Forbes has a very interesting column, building on some data from Bernard Munos (whose work on drug development will be familiar to readers of this blog). What he and his colleague Scott DeCarlo have done is conceptually simple: they've gone back over the last 15 years of financial statements from a bunch of major drug companies, and they've looked at how many drugs each company has gotten approved.

Over that long a span, things should even out a bit. There will be some spending which won't show up in the count, that took place on drugs that got approved during the earlier part that span, but (on the back end) there's spending on drugs in there that haven't made it to market yet, too. What do the numbers look like? Hideous. Appalling. Unsustainable.

AstraZeneca, for example, got 5 drugs on the market during this time span, the worst performance on this list, and thus spent spent nearly $12 billion dollars per drug. No wonder they're in the shape they're in. GSK, Sanofi, Roche, and Pfizer all spent in the range of $8 billion per approved drug. Amgen did things the cheapest by this measure, 9 drugs approved at about 3.7 billion per drug.

Now, there are several things to keep in mind about these numbers. First - and I know that I'm going to hear about this from some people - you might assume that different companies are putting different things under the banner of R&D for accounting purposes. But there's a limit to how much of that you can do. Remember, there's a separate sales and marketing budget, too, of course, and people never get tired of pointing out that it's even larger than the R&D one. So how inflated can these figures be? Second, how can these numbers jibe with the 800-million-per-new-drug (recently revised to $1 billion), much less with the $43 million per new drug figure (from Light and Warburton) that was making the rounds a few months ago?

Well, I tried to dispose of that last figure at the time. It's nonsense, and if it were true, people would be lining up to start drug companies (and other people would be throwing money at them to help). Meanwhile, the drug companies that already exist wouldn't be frantically firing thousands of people and selling their lab equipment at auction. Which they are. But what about that other estimate, the Tufts/diMasi one? What's the difference?

As Herper rightly says, the biggest factor is failure. The Tufts estimate is for the costs racked up by one drug making it through. But looking at the whole R&D spend, you can see how money is being spent for all the stuff that doesn't get through. And as I and many of the other readers of this blog can testify, there's an awful lot of it. I'm now in my 23rd year of working in this industry, and nothing I've touched has ever made it to market yet. If someone wins $500 from a dollar slot machine, the proper way to figure the costs is to see how many dollars, total, they had to pump into the thing before they won - not just to figure that they spent $1 to win. (Unless, of course, they just sat down, and in this business we don't exactly have that option).

No, these figures really show you why the drug business is in the shape it's in. Look at those numbers, and look at how much a successful drug brings in, and you can see that these things don't always do a very good job of adding up. That's with the expenses doing nothing but rising, and the success rate for drug discovery going in the other direction, too. No one should be surprised that drug prices are rising under these conditions. The surprise is that there are still people out there trying to discover drugs.

Comments (62) + TrackBacks (0) | Category: Business and Markets | Drug Development | Drug Industry History | Drug Prices


1. wwjd on February 10, 2012 2:14 PM writes...

Part of the problem with the Forbes number is they are pricing per new drug approved. A lot of R&D is spent developing new indications and dosage forms. They obviously ignore new vaccines (a new Flu vaccine comes out every year) but lots of R&D money get spent on those. I doubt pharma would be pushing a numnber like 1 billion if the truth was nearer 8.

Permalink to Comment

2. Student on February 10, 2012 2:15 PM writes...

I don't think we can ever really get a good picture. Even with a set definition, there are always going to be tricks.
Maybe someone wants to hide R&D expenditures in different departments (classify support staff as Sales rather than Research or something of that nature) so investors (we know how much they hate R&D) don't see them "throwing away money."
Conversely, maybe someone wants to inflate the numbers to flash at the FDA or convince drive the price up for a licensing deal.

Permalink to Comment

3. DrugA on February 10, 2012 2:23 PM writes...

Do these numbers take into account acquisition costs when a company grows its pipeline by buying smaller companies? It seems it would have to, but there is then a question whether that reflects the real development cost (ie dollars spent) or an over-estimate of the value in the pipeline.

Permalink to Comment

4. anon on February 10, 2012 2:47 PM writes...

Slightly different question:

If a start-up company was developing a single drug from start to approval, does anyone know how much money it would cost to do that?

Permalink to Comment

5. anon the II on February 10, 2012 2:56 PM writes...

It seems to me that money spent on chemists was a very critical but not so large part of that R&D expenditure. And much of that expenditure sits in the institutional knowledge that has been discarded along with those instruments on Dove bid. Sad.

Permalink to Comment

6. Spike on February 10, 2012 2:56 PM writes...

At least Forbes are being efficient and utilizing one piece of data in multiple places. The cost of a Superbowl commercial was also published today by Forbes as being approximately $3.5M (without production costs). It is interesting that in a article that talks about the high cost of drug development that includes all costs (successes + failures + overheads) compares it to the cost of a Superbowl commercial, the total costs of which are underestimated.
Not sure how many commercials were aired but let's say 40 - that will have "cost" $140M. In terms of providing benefit to society, I guess that we look pretty darned efficient

Permalink to Comment

7. SwedenCalling on February 10, 2012 3:04 PM writes...

Thanks for your view and the link to Forbes Derek. Thoughtful and Interesting as always. I think this is pretty much like Leesons attempts to correlate simple descriptors like MW and logP to drugs on the market. There are simply too many unknowns to draw any real conclusions. Useless basically. But we do spend to much and lipophilicity is bad. That's for sure. Found another interesting piece on Forbes by Mike Myatt, about leadership styles. If only the guys ruling our world would be better at listening...

Permalink to Comment

8. Joe Corkery on February 10, 2012 3:19 PM writes...

Interesting data, but looking at a static window of time like this seems to be a slightly flawed methodology because for drugs approved early in the window, you will vastly underestimate the costs associated with them and for drugs that don't get approved until after the window, they will contribute to a significantly higher overall cost. Perhaps the hope is that they will all even out, but I'm not sure that is a good assumption. Perhaps that is another contributing reason as to why the numbers were so different for the different companies.

I'm not sure about the best way to do this, because you are always going to have multiple projects in the pipeline in an analysis like this, but it might be worth looking at the costs that went into developing a first drug for a company as a better indicator. For instance, has anyone looked at what it cost for Vertex to get amprenavir approved as a benchmark? And how about all the costs since up to the kalydeco and telaprevir? Ideally, these costs should go down per drug as R&D is amortized over time and projects, but I think that would be a more interesting study.

Permalink to Comment

9. RD on February 10, 2012 3:24 PM writes...

Did we force the MBAs to relocate everyone from the Midwest to the more expensive cost of living states on the coasts? Did we force them to merge company after company and put projects on hold for a couple of years, paying researchers to vamp while they rearranged the deck chairs? Did we force the MBAs to hire expensive consultants to make our jobs harder, make processes more cumbersome, let the corporate IT people have control of the research computers and in silico environment? How much money could we have saved if we hadn't been subjected to stupid management tricks?
And what about the FDA? Did they have a responsibility to modernize? Let's not even start on the expense and risk aversion caused by the out of control class action industry.
Finally, how much more quickly would we have gotten things done if we hadn't had to ration human resources all the damn time?
The industry has been mid train wreck for almost two decades now and a lot of this has been self-inflicted.

Permalink to Comment

10. pete on February 10, 2012 3:45 PM writes...

So if it's 4-12 x 10^9 USD/new drug, then we'd better rethink the movement toward smaller market, sub-population therapies.

Tenable business model only if such drugs are:
- unusually cheap to discover/develop (..doubt it)
- really successful (..hope springs eternal)
- sell for an unusually high price (..the GOP may help you here)

Permalink to Comment

11. HelicalZz on February 10, 2012 3:50 PM writes...

I made a couple of comments in the article itself (casalera). Some referred to the tax bias to allocating expenses to R&D which Derek mentions, and I agree doesn't explain away much. But also I wondered about M&A activity. When acquiring a drug in development, you effectively capitalize the R&D done elsewhere, and the full R&D spend doesn't get captured in the expense line.


Permalink to Comment

12. J. Peterson on February 10, 2012 4:02 PM writes...

I work in another industry, and always felt bad that the first half-dozen or so years of my career went towards product development that never shipped. But two+ decades?!

That's a whole new perspective.

Permalink to Comment

13. skatesailor on February 10, 2012 4:13 PM writes...

Mr. Herper does not present the correlation coefficients among his variables. They are negative 0.503 (number of drugs, R & D spending per drug), +0.430 (number of drugs, total R & D spending), and +0.386 (R & D spending per drug, total R & D spending). The respective t-statistics are negative 1.421, positive 1.506, and positive 1.323; with t-critical equal to plus or minus 2.23. At the 5% confidence level and with 10 degrees of freedom, none of three hypotheses can be rejected. They state that no relation exists between drug approvals and spending per drug, none between drug approvals and total spending, and none between spending per drug and total spending.

With alpha equal to 0.20 instead of 0.05, we can be somewhat confident that the number of drugs declines [sic] with increasing spending per drug, but rises with climbing total spending. In these cases, t-critical falls to plus or minus 1.37.

Permalink to Comment

14. NoDrugsNoJobs on February 10, 2012 5:36 PM writes...

I consider it almost impossible to make a drug these days - That is the reality. And if one is going after a new target or broad indication or whatever than that statement is even more true. Look, we can blame the stupid MBA's, the CEO's or whatever all we want but no matter how you slice it, we still spend a shitload of money on R&D. So what if there were less mergers, etc - even the companies that didn't merge aren't really doing a hell of a lot better, are they? (e.g. LLY). The fda doesn't want to approve drugs, the governments/insurers/people do not want to pay for drugs, and if you are successful, people will be lining up to sue you for taking your drug. What's the point? Using my own money, I'd be making widgets or sport's drinks or whatever...the return is more assured and the outlays and risks smaller.

Permalink to Comment

15. Biotechtranslated on February 10, 2012 5:38 PM writes...


There is no set number for what it costs to bring a drug to market. It can vary incredibly.

The biggest chunk of cash is required for development, not initial research. When I was in big pharma, the number people liked to throw around was 1/3 R and 2/3 D for how a typical R&D budget was split (and there are several fold more projects in R than in D at any given time).

If you really luck out and discover a novel lead that doesn't require much optimization (ADME stuff), it's for a life threatening orphan disease with no current way to treat it (FDA hurdle is low) and it has a relatively clean profile (no serious safety issues and some level of efficacy), I would not be surprised you could go from pre-clinical to NDA approval for less than $50M or so.

Unfortunately, those types of drugs only exist in scientists' wildest imaginations. :P


Permalink to Comment

16. JasonP on February 10, 2012 6:55 PM writes...

It seems like the real question is how do you reduce the cost of clinical development? There needs to be a development revolution of some form.

Or perhaps make all marketing of drugs illegal and create a centralized method of drug information promotion. That would free up half the damn budget of any large company, who are usually the closers of drugs in clinical development. Wouldn't the ethical thing to do be to let a drug promote itself by its data rather than how many TV ads you can spread around?

Permalink to Comment

17. Matthew Herper on February 10, 2012 7:51 PM writes...

OK. Here we go...

@1 wwjd: fair point, and fairer for some others. I'll do some more work. There are some companies who might see their average cost go down.

@3 DrugA: they don't take into account acquisition costs. (I did that with Pfizer several years ago, and the numbers were daunting!) But I did the calculation this way to help taking into account the effects of mergers, which temporarily boost R&D.

@4 anon: One example, given in the article, is Optimer, for $175M. I think of I know of a few that are cheaper. And there's Questcor, which is a special case -- they didn't invent their drug, they raised the price.

@10 Orphan drugs usually are a lot cheaper to discover, faster to test, and are often cost effective at higher prices. This model has some legs to it.

In answer to a lot of other questions: I've quoted Munos on the $4B number before, when I profiled him back in August. He says internal estimates for per drug costs at most companies are that high and rising. (So does Corey Goodman, who would know.)

I'm not going to swear by the estimate for each company. But I do think it is pretty clear that the cost per drug is in the billions. And I think we are starting to get signs of fundamental change, which could be a good thing.

Here's to new medicines that really help people!


Permalink to Comment

18. Biotechtranslated on February 10, 2012 7:59 PM writes...


Marketing is not included in the costs that were the focus of this paper, this is truly just R&D costs.

The problem with reforming clinical development is that you have to get it past the FDA! It's hard to reduce the costs of a 10,000 patient clinical trial when the FDA says you won't get approved without it.

And I'm not sure I agree with your comments about banning marketing of drugs. Pharma companies don't spend millions a year on promotion because it costs them money, they spent it because it makes them MORE money.


Permalink to Comment

19. quasimotto on February 10, 2012 9:39 PM writes...

Society must place a great value on life if we are willing to pay this kind of monetary price to extend and improve it. Regardless of the reasons for the cost, we are willing to pay.

Permalink to Comment

20. Shanedorf on February 10, 2012 10:29 PM writes...

We can certainly quibble over the cost calculations, but we might be missing the mark.So far all the focus on cost containment hasn't worked. Maybe a better way to reduce the cost of drug development is to reduce the time of drug development. I know it sounds naively simplistic, but imagine a document landing on your desk at 5 pm and you wait until the next morning to finalize it. That's 15 hours wasted. Multiply that times thousands of people moving (or not moving) thousands of documents/data/reports over a 12 year development program. That's an astronomical amount of time.

If a moderately successful drug nets $1 million /day and the patent clock is ticking - then the focus on reducing costs is missing the point. Reducing time is a better strategy imo. Documents, reports, data crunching, even sample analysis should all be happening around the globe with teams in Tokyo,London, Boston, SF, etc. We have the technology to squeeze significant time out of the process. Pharma is global, CRO's are global. After 20 years in the industry I find it astounding that any development tasks could sit idle in an internet connected world.
"We have met the Enemy and he is Us."

Permalink to Comment

21. pete on February 11, 2012 12:52 AM writes...

@17 Matt H.
Good point about orphan drugs.

However, I was thinking about moves to target subtypes of major disease. For example, going after a particular molecular phenotype seen in a cancer subpopulation. So, where the biology might be extremely compelling for use of drug-X in this group of patients, the overall disease incidence (population size) may be rather modest. And even if the host company can get approval for drug-X in 2-3 different cancer subtypes, the aggregate patient size may still be inadequate to offset drug development costs of the scale that you cite. What a gloomy scenario for patients and drug hunters alike.

Permalink to Comment

22. soWhat on February 11, 2012 6:45 AM writes...

@20: Interesting point, but all these time savings will be lost asnyway in the next reorganisation.
Motivation is another point, good individual or even site performance is no job guarantee. In the next disease area strategy you are fired either way.

Permalink to Comment

23. BossHogg on February 11, 2012 7:04 AM writes...

Derek or Matt,
There is always a lot of assumptions and fuzzy math on this subject. We know it costs a lot and big pharma takes a lot of critisism from public and politics over the cost of drugs. The insight I have always wanted to see demonstrated and which never gets any press, is how much of the retail drug cost is actually attributed to manufacturing/intellectual property of the drug vs. pharmacy and hospital markups. I get that everyone needs to make money, but it seems the burden of development and letigous risk lays with BigPharma. So why when I go to the ER do I have to "pay" $40 for an aspirin. And when you take the numbers in this article and subtract from the price of a generic since they have virtually no development risk or marketing costs, shouldn't generics be a hell of a lot cheaper than they are?

Permalink to Comment

24. Jon Nilsen on February 11, 2012 10:14 AM writes...

@18 Depends on what you consider marketing. Medical Liaisons and MedComm both fall under R&D and most wouldn't consider those true costs of drug development.

R&D spend also includes all post marketing studies, obligates or not; and again would arguably not fall under the cost of developing a drug.

The other elephant in the room that limits the usability of this data, is that it rests on the assumption of existing market forces that function properly to set the costs of drug development. As all companies set R&D spend as a percentage of revenues and ,as Derek mentioned, there are pretty hefty tax advantages to expensing something under R&D if the money is going to be spent anyway. All the numbers tell us is what the organizations were willing to spend on "R&D" expenses. Not surprising that the larger companies with higher revenue had higher costs to develop drugs, Wall Street Allowed them to throw more money at "R&D."

This analysis answers the question "How much did pharma companies spend to run their operations?" Not "How much does it cost to develop a drug?"

Permalink to Comment

25. MTK on February 11, 2012 10:35 AM writes...


sounds great, except that most of the time is not "wasted" sleeping.

The time sink is the clinical trials, which are always (or at least should be for any project) on the critical path. It's not too easy to shorten the time it takes to run a clinical trial.

Permalink to Comment

26. johnnyboy on February 11, 2012 11:12 AM writes...

Sniping about the methodology of Forbes' number gathering is academic and pointless - as Herper points out, the purpose of the exercise is not exactitude, it's to give a reasonable estimate, and personally i'm satisfied with its validity. But not so satisfied about what this means - as Derek says, I don't think I see how that level of spending could be sustainable. This makes me seriously wonder how long it's going to take before we all pack it in and go fishing. At some point, it's entirely possible, perhaps after a couple of the larger pharmas sell out or go bankrupt, that the message comes through that we have done all we possibly could in terms of developing new drugs, and that pushing forward simply will not provide benefits in relation to the staggering costs. How many billions is worth a couple of months of cancer-free survival ? Sure there are orphan indications that will be worth pursuing, but things like high blood pressure, cholesterol, metastatic cancer - when will we give up on those ?

Permalink to Comment

27. polymer bound on February 11, 2012 11:47 AM writes...

I feel like some of this cash could be saved by letting science dictate clinical trial design instead of marketing. I know of a few phase III trials that I either wouldn't have let happen or would have redesigned had I been the one signing the checks.

Permalink to Comment

28. Anonymous on February 11, 2012 12:04 PM writes...

I second johnnyboy. Well said.

Permalink to Comment

29. cirrus on February 11, 2012 2:31 PM writes...

Derek: you are in the wrong business. You are more productive writing this blog, which is very good, than doing drug discovery, which is very bad as you have admitted. Nothing is that simple as just adding things up. Besides just making drugs, the pharma industry produces many other benefits to the society. At least, you are making a living and raising a family out of it.

Permalink to Comment

30. TJMC on February 11, 2012 3:41 PM writes...

I have been looking at the metrics on Pharma R&D for over twenty years. The span of comments above reflect its complexity and some of the confounding issues. Some additional points (some of which may be a recap of above):

1)Understanding performance should begin by comparing apples to apples. NME approvals are not dollars and it can be misleading to represent it that way. A better (best?) way is to measure the net present value (NPV) of the approved drugs’ revenues against the NPV of aggregate R&D spending.

2)Some estimates could be applied to remove the tax avoidance and “what portion is for marketing of Phase IV” confounders, etc. However each firm has reasons for very differing allocations. Therefore, these approaches may be useful for overall industry measures but fail when trying to compare firms to one another (unless you have that internal data across all.)

3)So, there are plenty of statistics and analytics we can apply (and we keep trying – myself included.) But we end up misleading ourselves to at least some degree.

Beyond understanding these issues, what is our PURPOSE? To improve the performance of R&D and to what ends? To uncover those (past or future) winners and losers? To lambaste management or justify our POV?

One simple idea is to just look at what the industry has done overall in terms of surviving or growing. If it serves society well, society rewards it. The fact is that returns on the industry were in the high teens in the 1980s and 1990s (and maybe earlier.) The past decade has seen that drop in half. Maybe stabilized and maybe reversing that trend. The key is, what are we doing to change that decline? Can we do more of them or additional things (and they need to be new, not just six-sigma)? What firms are winning comparatively and can others follow suit?

Permalink to Comment

31. Pete on February 11, 2012 7:27 PM writes...

Matt, Thanks for the clarification with respect to DrugA's question about acqusitions. It makes the cost of AZ's acquisition of MedImmune, for what I believe was about $15.6 billion, even more terrifying than it was at the time.

SwedenCalling, Attempting to correlate correlate simple descriptors like MW and logP to drugs on the market is really vital work. Surely new drugs would cost even more if this important work was not done.

Permalink to Comment

32. Jose on February 12, 2012 1:22 AM writes...

"Seattle's Sage Bionetworks seeks a drug-discovery revolution
When Stephen Friend formed the nonprofit Sage Bionetworks, his aim was nothing short of revolutionizing the way drug discovery is done so patients don't have to wait for help."

Seriously? How many time have we been through this? Don't drink the Kool-Aid!

Permalink to Comment

33. Rich on February 12, 2012 5:19 AM writes...


The majority of the cost comes from failure. What's worse, it's mostly copied, joint failure. Pharma has to embrace joint, pre-competitive research up to some proof of concept point (like PII). The current competitive landscape is clearly completely unsustainable. The Sage network, in particular their ARCH2POCM set up, is