Cost to bring a drug to market: Is it really $802m?

In April 2006, I reviewed a well-known article which stated that the cost to bring a pharmaceutical to market was $802 million.  This included the the cost R&D, clinical trials, failed drugs, and the cost of capial.  The estimate was from DiMasi, Hansen and Grabowski (2003).  The $802m figure was an updated from DiMasi, Hansen, Grabowski, Lasagna (1991) which claimed the cost to bring a pharmaceutical to market was $231 million in 1987 dollars.

Are these numbers correct?

An article in the Harvard Health Policy Review is suspicious. First, the data used to calculated the cost of bringing a drug to market was confidential.  This makes independent verification of these figures impossible.  Confidentiality may have been necessary to convince the pharmaceutical companies to share their cost data with researchers, but without independent verification it will be difficult to acertain if this $802m figure is exactly correct.

Secondly, Light and Warburton’s critique claims that about half of the pharmaceutical companies invited to participate declined to do so.  Since no comparison is made between the pharmaceutical companies who decided to participate in the study and those that did not, it is unclear of the cost data are representative.

The authors also claim that tax subsides and tax deductions should not count towards the cost of the manufacturing the drug.  From the firm’s perspective this is true, but from a societal point of view, these are still costs that someone (the taxpayer) has to incur.  I think DiMasi et al. were wise to include these costs in their study.

So what is the actual cost to develop a pharmaceutical?  Regardless of the exact figure, the answer still is ‘a lot’.

1 Comment

  1. Dear Jason,

    Let me briefly comment on the two areas that you note for which Light and Warburton expressed suspicion regarding our study (verifiability of the data and representativeness of the sample firms). First, the study had a section in which a wide variety of published data were used to corroborate our results. Second, it is not true that no comparison was made between the firms that participated and the firms that declined to participate. We noted in the study that the therapeutic class distribution of the firms that participated in the study and the distribution for all firms (i.e., all firms that were asked to participate) were virtually identical. They also did not differ materially in aggregate in terms of success rates or development times (industry values, with their associated larger sample sizes, were therefore used in developing the cost estimates). We are not aware of other characteristics in which these sets of firms should differ materially with regard to cost estimates.

    Joe DiMasi

Leave a Reply

Your email address will not be published. Required fields are marked *