Monday, January 14, 2013

Bad Pharma: A Cherry Picked NON Evidence-based Critique of Pharma Marketing?

I finally finished reading the 400 plus page book Bad Pharma™ by BenGoldacre, a British physician, whose previous book, Bad Science, was a best seller in the "non-fiction" realm.

As the name suggests, Bad Pharma is a no-holds-barred critique of practically every aspect of the pharmaceutical industry from drug development to marketing. Chapter titles include reiterations of the "bad" theme such as "Bad Trials" and "Bad Regulators." The chapter on marketing, which comprises about 25% of the book, is merely titled "Marketing." Goldacre summarizes the raison-d'etre of pharma marketing as existing for "no other reason than to pervert evidence-based decision making in medicine" and proceeds to "take a look at this mysterious world."

Of course, pharma marketing is not "mysterious" to me. I am familiar with most of Goldacre's criticisms, having written about the same subjects for many years. I was surprised, therefore, when none of my posts to this blog was referenced by Goldacre!

For example, when Goldacre states unequivocally that "tens of billions of pounds are spent each year, $60 billion in the US alone, on medicines marketing," he might have referred to my post that disputed this number ("Promotion vs. R&Deja vu all over again!"; also, see my analysis below).

I'd like to focus on this issue again here because it illustrates how Goldacre "cherry picks" data to buttress his argument despite his claim otherwise ("this book is based on systematic reviews, which summarise all of the evidence ever collected on a given question") and his claim that the pharma industry does the same; e.g., fails to report negative trial data, which is the focus of the first chapter of his book, "Missing Data", and which is "key to the whole story."

How much money the drug industry spends on marketing vs. research and development is a contentious issue, as Goldacre points out. "Marketing spend is a contested area, as the industry is keen to play it down," says Goldacre in the Notes section of the book where he could have referenced my post above, but did not. Nevertheless, he summarizes the issue this way:
"A quarter of the pharmaceutical industry's revenue is spent on marketing, twice as much as it spends on research and development, and this all comes from your money, for your drugs. We pay 25% more than we need to, an enormous mark-ip in price, so that tens of billions of pounds con be spent every year producing material that actively confuses doctors and undermines evidence-based medicine." 
"This is a very odd state of affairs," says Goldacre.

It's also "odd" that Goldacre's math does not add up. The value of the global pharmaceutical market was estimated to be US$880 billion in 2011 according to IMS Health (see "Pharmerging Markets to Continue Strong Growth"). One quarter of that would be over US$200 billion. All of this, of course, is not spent in US marketing, which Goldacre pegged as US$60 billion. However, I cannot imagine that the difference (US$140 billion) is spent in other countries. US marketing spend must be about 75% of the total in line with the overall size of the US market compared to the rest of the world. Even if it's only 50% that would mean US$100 billion spent on pharma marketing in the US, which is quite a bit more than the US$60 billion that Goldacre cited a few pages earlier in his book.

OK, that's a math error. But Goldacre also cites conflicting numbers for pharma marketing spending in the US. While he cites US$60 billion on page 246, on page 317 he says "we know that $30-40 billion is spent by the industry on drug marketing in the America" (aka "US").

I don't want to nitpick, but which is it: $60, $100, or $30 billion? Obviously, whatever the number, it's big. But is it twice as big as the R&D spend? That's the main thing Goldacre and other critics of the drug industry want people to believe in order to blame marketing for the high cost of drugs.

Let me revisit my critique of the $60 billion number that Goldacre cited and thus debunk the notion that pharma spends twice as much on marketing as it does on R&D.

The study that Goldacre cited for the $60 billion estimate was done by two Canadian authors: "The Cost of Pushing Pills: A New Estimate of Pharmaceutical Promotion Expenditures in the United States," which was published in PLoS Medicine.

Their conclusion: The pharmaceutical industry spends almost twice as much on the marketing and promotion of drugs than on research and development (see the story here).
"For the last 50 years, say the authors, there has been an ongoing debate as to which image [see left] of the drug industry is most accurate. The industry promotes a vision of itself, say the authors, as 'research-driven, innovative, and life-saving,' but the industry's critics contend that the drug industry is based on 'market-driven profiteering.'"
To prove this, the authors use data from 2 different sources: IMS and CAM as shown in this table:


Is anyone troubled by the methodology of the authors?

I mean, what could be easier than cherry picking which data to use and to make sure to always use the highest estimate to prove your point, which you obviously believed was true before you even did the "analysis?"

In essence, the authors place more value on the CAM data then on the IMS data. I'm not sure why, since both companies provide services for a fee to the pharmaceutical industry. The main difference between the two company's methodology is that IMS surveys the industry whereas CAM surveys doctors.

If you ask me, one methodology (IMS's) overestimates R&D costs and underestimates promotional costs, while the other (CAM's) is prone to exactly the opposite.

The biggest differences between the datasets involve cost of samples, cost of detailing, and "unmonitored promotion."

Cost of Samples
There is a large difference between the IMS (US$15.9 billion) and CAM (US$6.3 billion) estimates for cost of samples. Essentially, IMS bases its estimate on the retail value of samples, whereas CAM bases it on the Average Wholesale Price.

Which number would you use?

The PLoS authors use the IMS retail cost estimate because that is how the drug industry itself estimates the value of free samples it gives away! This is just another case where the industry shoots itself in the foot. It's hard to argue with the PLoS authors' statement that "it is inconsistent for donations to be reported in terms of retail value and samples in terms of wholesale value."

Cost of Detailing
The authors point out that "There is a significant discrepancy between the two sets of data in the cost of detailing: US$7.3 billion for IMS and US$20.4 billion for CAM."

IMS only considers the "cost to field the rep" and doesn't include in its estimate--as CAM does--indirect costs like costs for the area and regional managers, the cost of the training, and the cost of detail aids such as brochures and advertising material.

The authors claim that "relying on physician-generated data to estimate the amount spent on detailing [which is CAM's data collection method] is likely to give a more accurate figure than using figures generated by surveying firms [which is IMS's method]. Companies may not report some types of detailing, for example, the use of sales representatives for illegal off-label promotion, whereas doctors are not likely to distinguish between on- and off-label promotion and would report all encounters with sales representatives."

I dunno. If it were up to me, I would use a number somewhere in between the IMS and CAM estimate--say $10 billion.

"unmonitored promotion"

I am especially troubled by the $14.4 billion estimate for "unmonitored promotion." All that the PLoS authors have to say about this is:
"We believe that it is appropriate to correct for unmonitored promotion and that the figure we used is a reliable estimate. The 30% correction factor is based on a direct comparison that CAM is able to make between the data it collects through its surveys and the amount reported by companies."
May I say that this hardly rises to the level of objective analysis of available data? My analysis would completely discount this number as something that cannot be known and therefore should not be used. Besides, the same could be said for R&D spending and the two estimates would cancel one another.

My "NEW NEW" Estimate
Instead of the estimate of $57.5 billion that the PLoS authors use for promotional spending in the US, my NEW NEW estimate is $32.7 billion, which is pretty close to the $29.6 billion the PLoS authors use for the R&D estimate. This would lead me to conclude that--within the sampling margin of error, the US pharmaceutical industry spends about as much on R&D as it does on promotion. There! Everyone's happy!

Of course, I don't expect Goldacre to be happy with my NEW NEW estimate. He is a doctor and therefore would tend to favor physicians' anecdotal estimates vs industry's estimates of how much is spent on pharma marketing. That is, he favors a distinctly NON evidence-based approach when it suits his goal.

What is your estimate of the ratio of global pharma spending on marketing vs R&D?
Marketing Spend = R&D Spend (about same)
Marketing Spend = 2 X R&D Spend (twice as much)
Marketing Spend = 0.5 X R&D Spend (about half)
I prefer to avoid the issue altogether
  

14 comments:

  1. Yep, he's a cherry-picker extraordinaire. The title of the book alone is enough to give pause. There's certainly no harm in critiquing the industry, as it certainly needs to be strongly evaluated and issues brought to light, and yes, called on the carpet and critiqued when necessary....but any book that wholeheartedly endorses "BAD" as the over-arching theme should be taken in that context...Evidence-based? I don't think so.

    ReplyDelete
  2. Hi there, Ben Goldacre here.

    I'm very happy to correct any errors or inconsistencies in the book - it would be bizarre if there were none - and am always glad of help on that front, but I don't think you have identified an example of cherry picking here.

    It's worth noting that there are both higher and lower estimates around for marketing as a proportion of sales. The World Health Organisation said that marketing was a third of all sales revenue, for example.
    http://www.who.int/trade/glossary/story073/en/index.html

    Your main concern seems to be the 2011 PLoS Medicine paper whose $60bn figure you disagree with.
    http://www.plosmedicine.org/article/info:doi/10.1371/journal.pmed.0050001

    That paper also discusses other estimates, which are consistent with theirs, and which give estimates of marketing spend from a quarter to a half of total sales revenue (my casual mention of "a quarter" is actually at the lowest end of this range):

    "Our new estimate of total promotion costs and promotion as a percentage of sales is broadly in line with estimates of promotional or marketing spending from other sources. The annual reports of Novartis distinguish “marketing” from “administration.” Marcia Angell extrapolates from this annual report to the entire industry and calculates a figure of US$54 billion spent on pharmaceutical promotion in the US in 2001 [22]. As a proportion of sales, she estimates 33% is spent on marketing. Using similar methodology, the Office of Technology Assessment derived an estimate for marketing costs in the US by extrapolating from the cost structure of Eli Lilly. The Office of Technology Assessment considers that firms spend around 22.5% of their sales on marketing [23]. Based on United Nations Industrial Development Organization estimates, a report from the Organization for Economic Cooperation and Development estimated that, in 1989, pharmaceutical firms globally spent 24% of their sales on marketing [24], but few details of the methodology used were provided, making it impossible to verify the accuracy of the estimate. Finally, in 2006 Consumers International surveyed 20 European pharmaceutical firms to obtain more information about their exact expenditures on drug promotion. Among the 20 firms contacted, only five agreed to provide separate figures for marketing, which ranged from 31% to 50% of sales depending on the firm [25]."

    "The results are also consistent with data on the share of revenue allocated to “marketing and administration” according to annual reports of large pharmaceutical companies, if we consider that the largest part of “marketing and administration” is devoted to promotion. Lauzon and Hasbani found that 33.1% of revenues was allocated to “marketing and administration” [9], similar to the 31% reported by the Centers for Medicare and Medicaid Services [26] and the 27% from Families USA [27]."

    I'll post on your analysis on the "quarter" estimate in a moment.

    ReplyDelete
  3. Most importantly, you aren't comparing like with like in your blog post above. You take the $60bn US marketing figures from 2004 (from the Lexchin paper) then try to demonstrate that this cannot possibly make a quarter of global revenue, by setting this 2004 marketing spend in context of sales revenue figures from 2011.

    There was huge expansion in pharmaceutical industry revenue between 2004 and 2011, from $441bn to $660bn according to the FDA. It's also worth noting that your figure for revenue in 2011 is $880bn, which is a full third higher than the FDA figure, and further exaggerates the effects of your incorrectly setting 2004 marketing spend in the context of 2011 revenue.
    http://www.fda.gov/AboutFDA/CentersOffices/OfficeofGlobalRegulatoryOperationsandPolicy/GlobalProductPathway/ucm259754.htm

    The $60bn marketing spend for 2004 should be interpreted in the context of global revenue in 2004. So, pursuing your "back of the envelope" theme, the marketing spend in the US alone in 2004 ($60bn) was around 15% of sales revenue in 2004 ($441bn). I don't think it's unimaginable that the whole of the rest of the world made up the difference to a quarter. And the very many studies described above, which estimate marketing spend at between a quarter and half of sales revenue, are likely to be more thorough than your back-of-the-envelope calculation, in which you jumble up 2004 and 2011 figures, and my back-of-the-envelope correction using 2004 figures.

    So, overall, the estimate of "a quarter" doesn't seem outlandish, or unreasonable, and it certainly isn't very unusual, nor is it anywhere near the highest estimate of 50%.

    That said - as I explain very openly in the book - I didn't want to go too large on marketing spend, over the course of 400 pages, since there are multiple conflicting estimates for it, and marketing isn't my main area of interest. So I'll happily have a look at those bits of the book and see if there's anything worth revising for the next edition.

    I should add, lastly: there are interesting problems and shortcomings throughout medicine. That is the nature of science. The book sets out these problems, and offers the public accessible explanations of things like trial design, pharmacovigilance, evidence synthesis, etc. It recognises the good that the pharmaceutical industry has done, and is clear that the individuals working in it are not bad people. It should be acceptable to discuss these problems with the public, without eliciting shrill denunciations. I hope that the public can help us to devise ways of improving the systems in which we operate, especially in areas of missing trials, where the problems have persisted unfixed for far too long, and I hope industry can engage in these discussions.

    Best wishes,

    Ben Goldacre
    www.badscience.net

    ReplyDelete
    Replies
    1. Ben,

      Thanks for your comments. All these numbers are making my head spin :-).

      Delete
  4. I'm sorry if these numbers are taxing for you. I have only posted because you actively solicited me to respond to your blog, which is obviously rather intemperate in its tone.

    It would be good to have your view on:

    1. Is it appropriate for you to interpret 2004 marketing figures in the context of 2011 revenue (and a very high estimate of 2011 revenue at that)?

    2. Is it really reasonable for you to accuse someone of cherrypicking "a quarter", which you claim is chosen to be unusually high, when in reality half a dozen references from one paper show the range to be from a quarter to a half, and WHO cite a third?

    Best wishes,

    Ben Goldacre
    www.badscience.net

    ReplyDelete
    Replies
    1. No problem. I was just making a little joke instead of taking time to review your comments in detail. I do appreciate your comments and will review them more thoroughly.

      Regarding "cherrypicking," I think I was only referring to your selection of $60 billion from the PLoS study vs. the lower IMS number.

      The percent of sales that this spending represents is something I haven't thought about much, so I will have to look more closely at the estimates you provided to see if I agree or not with 1/4, 1/2, or 1/3. Considering the huge numbers involved, your 1/4 estimate seems more reasonable than the others.

      But, I'm more interested marketing spend is the expenditure on R&D. I inserted the little poll to get some anonymous feedback regarding that.

      Delete
    2. Those are pretty grand and melodramatic accusations about cherrypicking that you make in your headline, given the holes in your back-of-an-envelope calculation, and the lack of evidence for cherrypicking.

      I don't think the "quarter" figure is cherrypicking, estimates range from a quarter to a half, maybe there will be some lower, but I didn't say "a half", I took the lower end, not the higher.

      In giving the $60bn figure I cite a paper and am candid in saying there are lots of estimates in this and I'm not very wedded to one. As you point out, elsewhere I give a different estimate, $30-40bn. I'll have a look at what to do about that in the text for the next edition, but that really doesn't look like cherrypicking to me.

      There are a lot of really serious issues here, that effect patient safety, this is not a deranged conspiracy theory book, it's a straight explanation for the public about how evidence based medicine works, and some barriers to it achieving its aspirations, not limited to industry, as you will know, from reading it.

      I honestly think it would be good for patients and industry if people could engage in a sensible conversation about this stuff without resorting to this kind of football team mentality or - with the best will in the world - shrill and rather overstated denunciations.

      Delete
    3. I think the title of my post is less accusative than the title of your book -- let's be honest: both are meant to "sell" the contents. At least my title included a question mark to entice debate, which is my goal.

      Delete
  5. Interesting discussion. Allow me to share a few observations.

    First, the big picture. At the heart of this discussion is the question of whether the pharma industry spends more on marketing than on research. There is an assumption behind that question, of course: we are supposed to believe that the pharma industry are not really serious about research if they spend more on marketing than they do on research.

    In effect, the assumption is that spending on marketing and research is a zero sum game. If you spend a dollar on marketing, then you have one dollar less to spend on research.

    The problem is that that assumption really doesn't stand up to critical thought. It's not a zero sum game. The whole point of marketing (at least if you're doing it right) is to increase revenue. If you spend money wisely on marketing, then you will have greater total revenue, and if you are spending a certain percentage of your revenue on research, then by spending more money on marketing, you will have more money to spend on research.

    There's an interesting blogpost that goes into more detail on that point here:
    http://blogs.discovermagazine.com/crux/2012/08/16/are-drug-companies-faking-an-innovation-crisis-uh-no

    Now, to look more at the detail, I can't claim any expertise in how much pharma spends on marketing, so I'll leave John & Ben to slug that one out among themselves. However, I do think John makes a good point about the way the numbers are presented in the book. It's pretty clear to me that there is considerable uncertainty about the numbers, depending on how you measure things. So for Ben to simply present these numbers in the book as established fact, without making all the complexities and uncertainties clear to the reader, does strike me as misleading.

    If we really want to calculate the ratio of marketing spend to research spend (although I'm really not sure there's a lot of value in doing so), then we also need to consider that research spend is incredibly hard to measure as well. These days, big pharma companies do not do all their own research themselves. A lot of research is done in tiny little startup companies, whose research will not figure in any accounts of research spending by big pharma. Big pharma then pay for that research by licensing deals or by buying the whole company. Figuring out exactly who's spent what on research under those circumstances is really not a trivial accounting problem.

    I also have to agree entirely with Ben when he says "it would be good for patients and industry if people could engage in a sensible conversation about this stuff without resorting to this kind of football team mentality". We're all on the same side!

    So it's a shame that so much of Ben's book seems to have been written with that "football team mentality". There are many places where he could have acknowledged the good work done by the pharmaceutical industry, but chose to take a pop at them instead. For example, the section on ghostwriting completely fails to acknowledge any of the efforts that the industry have made towards addressing this problem (other than by incorrectly claiming that any efforts have been unsuccessful), and didn't cite any of the evidence that ghostwriting is becoming less common. Would not a more constructive approach have been to criticise where appropriate, but also to bring in some balance by acknowledging the good stuff?

    For what it's worth, I've written some of my thoughts on Bad Pharma here:
    http://dianthus.co.uk/ben-goldacres-bad-pharma

    ReplyDelete
    Replies
    1. Fascinating discussion. It has taken a long time for this important issue to reach such a level of public awareness and we have Ben Goldacre's sincere efforts and provocative style to thank for that. I have also read his book and was indeed intrigued by the sections on pharma marketing spend and how he arrived at the figures presented.

      I have been involved in measuring pharma marketing spend since 1998 - first with the CAM Group and then with Cegedim subsequent to its acquisition of CAM - so this matter is of great interest and at the heart of my professional experience.

      From this perspective I must agree unreservedly with Adam Jacobs' comments, to wit: "there is considerable uncertainty about the numbers, depending on how you measure things. Witness our recent exchange concerning how to cost spending on samples. While Cegedim has assumed a cost based on average wholesaler price, another agency assumes that each sample drop is worth retail price. Both provide useful measures depending on the objectives of the analysis. Neither is "right" or "wrong". Over many years CAM/Cegedim have undertaken to validate our methodology and assumptions and the the industry has broadly come to accept our approach as a consistent means to benchmark promotional levels.

      Both Cegedim and the other leading agency with expertise in this area arrive at very similar levels of global promotional investment - over 90 Billion USD per year - roughly 10% of global sales. That being said, neither agency can claim to measure all promotional channels, so it is safe to say that 10% of sales is very likely an underestimation. Indeed, we can be quite sure the industry spends more than 10% of sales on marketing activity. Precisely how much is probably as impossible to determine as is R&D spend.

      I believe this is a healthy, necessary discussion. As it brings into play health, capitalism, ethics and a few other "footballs" it will be impossible to avoid the gridiron. So gentlemen, enjoy the game!

      ...and for what it's worth from time to time, check our my blog on news and comments around phamrma marketing http://pharmacw.blogspot.fr/

      Delete
  6. This comment has been removed by the author.

    ReplyDelete
  7. Interesting thread, although I actually think the % spend on marketing itself is not of great consequence. The implicit assumption seems to be Pharma Research = potentially good and Pharma Marketing = surely bad. I actually believe Pharma Marketing spend can be valuable to patients, HCPs and wider society if it is done in a balanced way, e.g. promoting a product that has value in a particular subset of patients, or raising awareness of a relatively undiagnosed condition.
    It is not HOW MUCH you spend rather HOW you spend your marketing budget. Even in a world of perfect numbers, you can't say for example, company A spending 35% on marketing must be therefore be using this fat to create dodgy data and inducements to prescribe. Company B only spends 10% so must be very focussed on only providing the minimum amount of balanced information and can't be doing anything unethical. Statistics rarely correlate - yet even if they did a deeper exploration of why is required.

    ReplyDelete
  8. Goldacre is certainly a cherry-picker. I quote, 'In the unlikely event that any example I have used is simply wrong, there will be another to slot in its place,' Bad Pharma P372, an attitude that the much maligned (by Goldacre) FDA will not tolerate in sponsors. In fact his analysis of problems in publishing is also wrong, although it has to be conceded that here he is (mainly) simply repeating the mistakes of others. See http://errorstatistics.com/2013/01/06/guest-post-bad-pharma-s-senn/ for more details and also http://f1000research.com/articles/2-17/v1 .

    My declaration of interest is here http://www.senns.demon.co.uk/Declaration_Interest.htm .

    ReplyDelete
  9. The bigger picture should be that pharma companies have changed. They now make medicine and their end goal is to have as many people on it as possible. It was once a goal for pharma companies to make a medicine and get it to those who need it.

    You continue to hear over and over about how side effects are downplayed, or after a drug has been released they find major side effects they hide under the rug. Telling doctors that certain antibiotics are good for everything, when they should be last to be considered for use due to it's strenthen and side effects it can have.

    Get over it. The companies need to clean up the act. Just look at J&J and their hip replacements ordeal or their LEvaquin ordeal. These CEO's should be thrown in jail for life for hiding terrible side effects and letting doctors prescribe these dangerous meds that should be left for last resort use.

    ReplyDelete