Health News Review

The following is a guest post by Chad Parmet, a Research Associate with the Informed Medical Decisions Foundation.  Chad gathers, evaluates and summarizes current medical research in support of new or revised Shared Decision-Making® programs. Before joining the Foundation, he worked as a medical writer. He holds a BA in physics from the University of Pennsylvania. In the past he reviewed many stories for (The Foundation funds this project but has no control over what is published herein.)


Often I read that researchers have cooked up some new miracle drug that will change the world. But many of these miracles never materialize. That’s why one of the criteria rates articles on whether they’ve set appropriate expectations about the availability of the hot new thing.

But what should our expectations be for a drug that is still being tested?

Before a compound can make it onto a prescription pad, it must run a gauntlet of research studies in animals and humans. The FDA reviews the results and decides whether to approve it for use in Americans.

So I’ve been wondering: What are the odds that a drug in the middle of that gauntlet, generating headlines along the way, will ever make it our medicine cabinets?

I poked around PubMed, Google, and TRIP for evidence about those odds. I didn’t see any meta-analyses. I found a number of studies that used different estimates to produce the odds. Clearly, I can’t get a confident estimate of the odds without finding a solid, systematic meta-analysis of high-quality studies. That said, the estimates were in the same ballpark. And the ballpark was interesting.

The graphic below is my “gist” of the results from recent, peer-reviewed studies that—in my opinion—were most generalizable to small-molecule drugs seeking their first indications from the FDA.

Odds that a drug entering each phase of research
will ultimately make it to the market


The gist of Adams & Bratner 2006, Davis et al 2011, DiMasi et al 2010, Kola & Landis 2004 and Paul et al 2010


Here are some important caveats about those numbers:

I could go on. This isn’t an academic study, just an exercise.

With these limitations in mind, the true gist I take from the exercise is that experimental drugs seem to have a lower likelihood of making it to the market than what I would expect from the revolutionary tone used in the typical news coverage of the next miracle drug.

I’d love to know if there’s a formal meta-analysis, or a better summary of the literature, that I missed. If you know of one, please leave a comment below.

Chad Parmet

Disclaimer: All of the research and conclusions presented are my own and do not represent the views and opinions of my employer, the
Informed Medical Decisions Foundation.



Greg Pawelski posted on July 17, 2012 at 1:19 pm

As you say, “this isn’t an academic study, just an exercise.” At a 2009 ASCO breast cancer symposium in San Francisco, the Keynote Address by Martine Piccart-Gebhart of the Jules Bordet Institute made a point that only 8% of new drugs entering Phase I trials ever make it to marketing and that this percentage is even lower for cancer drugs, and Chas Bountra of the Structural Genomic Consortium raised the sobering fact that 90% of Phase II compounds fail, because current drug testing is inefficient, with many drugs failing late in development, with these expensive failures owing, in large measure, to ineffective drugs and poor patient selection (i.e. lack of prognostic and predictive markers for response to therapy). Piccart-Gebhart went on to note that little progress has been made in identifying which therapeutic strategies are likely to be effective for individual patients. She concluded that identifying markers that can predict response to a particular drug remains a great challenge.

Musa Mayer posted on August 3, 2012 at 11:47 am

I’ve heard the 8% figure that Greg Pawelski cites as well, presented a few years ago by Janet Woodcock of FDA, who I guess ought to know. In fact FDA would likely be the only reliable source for such a number, I would think, since they must approve all such trials, and the law does not require listing of Phase I drug development trials in Since this is proprietary information, and the vast majority of these Phase I trials (those that fail) will go unpresented and unpublished, I think it’s pretty safe to assume that we will never know the denominator.

Chad Parmet posted on August 6, 2012 at 9:44 am

Thanks for the insights and suggestions, Greg and Musa. Very much appreciated. I will check out Dr. Bountra’s estimate.

FDA CDER told me the agency doesn’t collect the data I was looking for. I did see the 8% figure in FDA reports and presentations (e.g., but didn’t include it in this post because it cited estimates from Bain & Co (, Exhibit 2) that didn’t seem peer reviewed, which was one of my criteria. Perhaps it was peer reviewed elsewhere and I missed it? But it does seem notable that FDA deemed these estimates good enough to cite.