Yesterday Kelly Sexton tweeted out the following claim:
It would be great to think that the Bayh-Dole Act has resulted in a bunch of new therapeutics reaching the public. And perhaps that’s the case. And it might even be simpler to celebrate 300 new therapeutics in 35 years–those might be really important therapeutics, regardless of whether Bayh-Dole had anything to do with their success. But Sexton couches her report of happy outcomes with some claims worth considering. She claims “no therapeutics to market from University research” before Bayh-Dole. She claims that after Bayh-Dole, there have been 300. And behind these two claims is the implication that Bayh-Dole is the reason for the difference. And behind that is the implication that Bayh-Dole has been hugely successful. Let’s look at these assertions and implications.
Let’s start with therapeutics before Bayh-Dole. This part is easy. There were of course therapeutics arising from university research getting “to market.” We can start with heparin (1916, Johns Hopkins, into clinical trials in the 1930s), insulin (1921-22, University of Toronto), and warfarin (1930s to 1950s, University of Wisconsin, into clinical trials in the 1950s). (Here’s a convenient article on both heparin and warfarin). There are plenty of other examples. Cisplatin (1960s to 1970s, Michigan State, clinical trials 1970s), for one. In England, we have penicillin (1920s to 1940s, St. Mary’s Hospital (and medical school) and Oxford, in widespread use 1940s). From Austria, sulfanilamide (1908, Technische Hochschule, in clinical use 1930s). Streptomycin (1943, Rutgers, 1940s–Robert Dole was the first patient treated to survive its use without awful side effects).
Examples here are enough to show Sexton’s assertion isn’t true. There were university-discovered therapeutics reaching the market well before Bayh-Dole. Bayh-Dole wasn’t passed because university-discovered drugs were not reaching the market. It was passed, in part, because university-discovered drugs were reaching the market. Sexton has it backwards. But that’s not all.
Let’s look at some history, and then you can see how empty Sexton’s argument is with regard to Bayh-Dole. University patent-based work jump started with Cottrell’s creation of Research Corporation in 1912 to manage a portion of his patent rights for the electrostatic precipitator, which he had invented at the University of California. Research Corporation would accept faculty inventions for management and return a portion of any licensing revenue to the inventors or to the university (depending on how the university’s patent policy played out). Research Corporation used the remainder after costs to fund research nationally, first through the Smithsonian Institution and later directly. In 1925, Steenbock and others created the Wisconsin Alumni Research Foundation, which aimed to be a Research Corporation but specific to the University of Wisconsin. Many other universities followed WARF’s model and created an external affiliated foundation to manage patent rights. By 1962, Archie Palmer counted 55 such foundations at universities. Only a few universities–MIT, University of California, and Stanford among them–ran their own licensing operations. Most outsourced to someone, whether Research Corporation, Battelle, University Patents, or their own affiliated research foundation.
Getting an invention to a licensing agent was relatively simple. The inventor reported the invention to the university and the university cut a deal with regard to an equitable interest (if any) based on the circumstances of support and directed the inventor to an invention management agent. If the agent took the case, then the inventor assigned the invention to the agent, the agent covered the costs of filing a patent application, and if anything came of it, the agent shared royalties with the inventor and the university. The university administration was out the cost of the review and referral, but did not have to deal with the complexities of patent prosecution, license negotiation, contract management, infringement, or money. It was 99.9% upside or nothing. Pretty smart, if you think about it. Just take an equitable share of the successful stuff.
The problem arose as the federal government ramped up research at universities in the years following World War II. A question arose regarding inventions made with federal support. If that support was in the form of a procurement contract, then the federal government could require assignment of title to any inventions along with other research deliverables. If the federal support was more like an assistance grant to the university for general support or was more like a fellowship or other personal award made to a faculty member or student, then there ought to be no federal expectation of title to any invention. The problem was that federal funding from various agencies, and within those agencies, could shift between procurement and subvention, between (as Charles Kidd has it) purchase and support. And with the shift, patent rights could shift from the investigator (and university, depending on the policy) to the federal agency providing the funding. In the later 1950s, agencies adopted the idea of a “project” for university work, giving rise to “sponsored projects offices” to manage them.
“Project” shifted attention from the university or investigator using resources to follow a line of investigation to proposed work to be performed. In a project, the outputs follow from the proposal. For a project at a university, the federal government contracts with a university for the services of faculty (and other) investigators, which the university releases from other duties (often originally, in the summer, and later teaching). The investigators then are working for themselves (their proposals), for the university (which manages the federal money), and for the federal government (the sponsor expecting to benefit from the results).
The project was a compromise that made government funding administratively uniform but made a mess of most everything else. The government could expect to have rights of some sort in any invention, whether just a license or title or the right to place inventions in the public domain. By 1959, “project” had become the standard form of government funding concept–bits of purchasing but with a drift toward support. With project came the idea that the statement of work–the proposal–was the thing that was reviewed for support. Much depended, however, on the purpose for which a federal agency contracted research work–to meet agency mission or to enhance the general condition of science and scientific training? From the concept of project we get relatively directly to Bayh-Dole’s “performance of work under a federal funding agreement.”
In 1963, President Kennedy issued a Statement of Government Patent Policy that provided guidance on typical situations in which the government ought to require assignment of inventions, allow the contractor to own inventions (subject to a government license), or decide on a case-by-case basis whatever appeared to be in the public interest. Within the logic of Kennedy’s policy, universities ended up as contractors that did not have a clear-cut commercial position under which federal agencies could cede their claim to title. Thus, cases had to be dealt with as they arose. To navigate the Kennedy patent policy (and HEW regulations), in 1968 Norman Latker at the NIH, along with Howard Bremer at WARF, came up with the idea of a master agreement, the Institutional Patent Agreement, that would deal with case-by-case review for NIH work up front. The IPA introduced a number of novelties–universities had to require assignment of inventions made with NIH support, but only if the university had decided to file a patent application. Universities who obtained title to a subject invention had to show development of an invention to the “point of practical application” within three years of a patent issuing (according to the Kennedy patent policy); the government could otherwise intervene if the invention was not made available on non-exclusive terms or the university (and its partners) could not make a compelling argument that they should keep their exclusive position. Much of the IPA text, with subtle and significant differences in effect, ends up in Bayh-Dole.
By 1980, according to Bremer’s notes, over 70 institutions–(38 universities and university systems) had IPAs with the NIH (and some with the NSF). That is, by 1980, many universities doing the bulk of NIH research had a deal with the NIH that allowed the university to have clear ownership of inventions made with NIH support. The university could license exclusively, but had to limit exclusivity (according to the IPA) to three years from the date of first commercial sale or eight years overall, whichever came sooner). It was the network of private invention management agents and the IPA access to federally funded NIH inventions that provided the argument for Bayh-Dole making federally supported inventions generally available to the university-based network.
Perhaps now you see the problem with Sexton’s assertion about Bayh-Dole. Whatever the law did, it didn’t do it in a significant way for therapeutics discovered at universities with federal funding because most of this work already was covered by IPA agreements or could be otherwise released to university patent agents on request (but on a case-by-case basis, with the prospect for delays or refusals). It was this “uncertainty of title” that Latker and Bremer sought to deal with by creating from the IPA what became Bayh-Dole. Bayh-Dole allowed universities to claim any invention they wanted (subject to their own policies and practices with regard to claiming inventions–they didn’t have to claim anything at all, just as they didn’t under the IPAs).
Rather than look for an increase in university patenting after 1981, one should look for a decrease in university patenting in the period 1978 to 1981, between the end of the NIH IPA program and the effective start of Bayh-Dole. It’s just that if one looks, there’s no decrease at all in 1981-84, when patent applications filed in 1979-81 would have likely 
issued as U.S. patents. (The graph is based on my working through USPTO records.) If anything, nonprofit patenting of inventions made with federal support picked up in 1978-81, when the IPA program was shut down and Bayh-Dole was not yet in effect. The jump happened without Bayh-Dole. Other factors must be at work. Certainly not either the IPA program or Bayh-Dole. The big leap comes around 1992–meaning, patent applications filed around 1989, inventing taking place even a bit earlier, in 1988 and 1989. What were the factors that caused that jump in patenting?
Selectivity appears to be one of the secrets to the success of the private university invention management network: it was multiply selective. Investigators decided to seek patents, universities (committees first, then administrators working under policies) agreed that patenting was appropriate, and an invention management agent reviewed and agreed to accept an invention. A university did not claim an invention under an IPA (or bother to request that the government release its claim to require assignment of title) if the university could not find an invention management agent that would take on the case (or find a company ready to take a license and pay all the costs of patenting).
Bayh-Dole had next to nothing to do with giving universities any advantage with therapeutics discovered at a university with federal support. The IPA system under the Kennedy patent policy (as modified here and there by later executive orders) provided universities with all the access to those inventions that they needed. Whatever effects Bayh-Dole may have had on the growth of university-based patent-and-license operations, or on the number of inventions patented, or on the expansion of the IPA system to other areas of federally supported research, therapeutics is the area that experienced the least change as a result of Bayh-Dole. Again, Sexton has things entirely backwards. Bayh-Dole was premised on the apparent success of the university-based largely private invention management infrastructure. That infrastructure already had access to NIH-supported inventions made at many universities, if not through an IPA, then upon request to the NIH, which had led the creation of the IPA system.
The effects of Bayh-Dole were to disrupt the practices of some other federal agencies–notably NASA and the (relatively new at the time) Department of Energy–and force them to adopt (with exceptions) the “uniform” approach of Bayh-Dole. It is in these areas that one might then look for substantial changes in the disposition of federally supported inventions. This is what makes PhRMA’s arguments for the “success” of Bayh-Dole so nonsensical. Bayh-Dole didn’t significantly change NIH invention policy for universities that already had IPAs. It changed other things, but not this area.
Bayh-Dole did some things that Sexton won’t talk about. Bayh-Dole destroyed the prior invention management infrastructure that was doing so well (apparently, if the claims were to be believed) in comparison to the government’s own efforts. The argument was that Bayh-Dole would send more federally supported inventions into that private invention management system. Instead, university administrators working under Bayh-Dole destroyed that system and replaced it with university-based patent administration. That’s the growth of the university OTTs and TLOs, replacing national agents and shutting down or changing the role of many of the university-affiliated research foundations. And with the faux version of Bayh-Dole–that Bayh-Dole was a vesting statute, so universities owned inventions outright or nearly so–came the loss of selectivity. This loss of selectivity leads directly to the massive build up of patents held by universities. The growth in university patenting after Bayh-Dole is not a measure of increased productivity or innovation or even the potential for innovation. It is a measure of the loss of selectivity; the shift of academic budgets to pay for patenting, licensing, and litigation; and the shift of inventions from the private network (where patents were owned by the agent) to the universities.
The build-up over the past 35 years of unworked inventions controlled by unworked patents (unlicensed, licensed exclusively but not worked) is a measure of the amount of research from universities that has been cut off from the public domain, from immediate availability for use in research and company settings. We have had a huge loss in immediacy, in ease of access to inventions made at universities. Where before a publication might teach an invention and anyone who could read could practice, now practice requires first the negotiation of a 30-page patent license, payments, fuss about liability and governing law, and did I mention payments? The effect of Bayh-Dole is to delay by two decades the public availability of most of the significant findings of federally supported research. As university administrators extended what they took to be Bayh-Dole’s requirements to all inventions, and then even to non-inventions, they stoppered up much university research for public use. That effect will not be found in success stories. It is found in the absence of things. To look for absences–that’s way more difficult unless you know what should be there, but isn’t.
What’s remarkable is that for inventions made under the IPA system, the commercialization rate was about 4%–just what the government’s rate was, according to the advocates for Bayh-Dole. Just about what the historical average for the commercialization of patents is in general. The overall university commercialization rate by 1980, again according to Bayh-Dole advocates, was more like 25% to 30%. (Interestingly, according to Rebecca Eisenberg, the rate at the time for government licensing of biomedical patents was 23%–something Bayh-Dole advocates don’t like to mention). Again, the higher rate for university licensing appears to track the effect of multiple layers of selectivity before there’s ever a patent. The rate of commercialization now for university inventions appears to be less than 0.1%–two orders of magnitude less efficient, or two orders of magnitude more stuff held back from common use.
Bayh-Dole did absolutely nothing to change the patent situation for therapeutics discovered at universities. Bayh-Dole made a mess of things for everyone else. Bayh-Dole disrupted university patent policies, upset government oversight of patent administration, interfered with federal agency missions, hammered immediacy of access, and hammered selectivity of inventions suited for patenting. These are things worth civic discussion and bound any exclamations of the “success” of Bayh-Dole.
Now as for Sexton’s claim for 300 therapeutics developed after Bayh-Dole (and apparently because of it), it would help to have the list and the method. A link is all that’s needed in Twitter. A 2011 Boston University study looked at university-developed drugs made with federal government funding from 1970 to 2009 and counted 153. AUTM’s own 2015 “Briefing Book” cites the same number. Now there can be drugs developed at universities that didn’t have federal funding, and therapeutics that aren’t drugs, and of course there could have been a huge spat of new therapeutics from 2009 to 2016, but on the face of it, Sexton’s 300 looks high, perhaps 2x reality. Maybe she means any therapeutic for which a university contributed science or clinical testing or some variation. But if we’re talking Bayh-Dole, then there ought to be a federally funded university grant for which there’s a patent on a subject invention.
If we look at the FDA’s historic figures for New Drug Application approvals, we don’t see any Bayh-Dole effect. There’s no obvious uptick after 1981. I’ve circled 1980-81 on the chart. 
Any upticking was happening before Bayh-Dole. From 1950 to 1980, NDA approvals averaged 150 per year. After 1980, it was 90 per year. If there’s an effect from Bayh-Dole it is swamped out by other factors.
Bayh-Dole moved what would have been federally owned patents and patents owned by invention management agents to universities, but Bayh-Dole does not appear to have done anything to bring about significantly more NDA approvals. Even if we are talking 300 NDAs over 35 years, that’s still only about 10% of the total, where it appears that the federal government funds (estimates here) about 28% of the total R&D for medical research. There is still value in 300 new therapeutics, even if it is more like 153. It’s just that in context, Bayh-Dole doesn’t appear to have done anything positive worth remarking on, let alone a concerted effort to push the same “success” talking points onto the public (and to lawmakers). If Bayh-Dole is so wildly successful, why do the people who so promote it appear to be so dull, bureaucratic, and dedicated to deception and spin? If Bayh-Dole can’t survive an honest reckoning, why on earth should it survive at all?
Where do we end up, then, with Sexton’s Twitter post?
Universities were active in discovering therapeutics that reached market well before Bayh-Dole. So she’s wrong there.
There was a well established private network of invention management agents that university inventors could use when needed. That network was selective. Everything else was made available with immediacy and low administrative overhead.
There’s nothing to indicate that Bayh-Dole changed a thing for therapeutics developed with federal support at universities. That’s the one thing that Bayh-Dole did not significantly change. So she’s wrong there, too.
And it’s not at all clear that there were ever 300 therapeutics, or that they were a result of Bayh-Dole, or that 300 (or 153) should meet or exceed our expectations for what federal funding at universities should be producing. And of course, there’s no indication from AUTM folks about how many of these therapeutics are available to the public on reasonable terms. It would be great to get more from AUTM on their list of 300, how they determined university involvement and federal support, and the like. It’s not like a Twitter post can do much of this (though there are links!). But even a Twitter post can get things a whole lot more right than Sexton does.
It would help if AUTM leaders would show some awareness of history. There are stories to tell involving Bayh-Dole. Even happy stories. But a good start would be to work even a bit with the truth. That’s a problem for AUTM these days, and a problem for the public universities that pay for their patent administrators to be members of AUTM.
[updated May 2, 2017 to add patent graph and correct the technical details of how the IPA differed from the Kennedy patent policy (which the IPA should have followed but didn’t)]