An ARPA for Health Research?

In what would be the most significant change to US biomedical research funding policy in decades, Congress is considering a proposal to create an Advanced Research Projects Agency for Health (ARPA-H). As the name indicates, ARPA-H draws inspiration from the Defense Advanced Research Projects Agency (DARPA), which was created in 1958 by the Department of Defense in response to the Soviet launch of Sputnik and to worries about the loss of American technological leadership. DARPA developed a range of important military technologies and contributed to innovations that eventually had large consumer markets, including the laser, the Internet, robotics, autonomous vehicles, and GPS technology. It also helped create the mRNA platform for vaccine development. The DARPA model is very different from most other R&D funding efforts in the US government, and its successes have prompted thinking about similar approaches in numerous sectors beyond defense, including for energy and climate solutions. Though its essential features have evolved over time, the approach involves identifying specific bold goals, funding and coordinating the R&D needed to achieve them, and implementing solutions at scale. A core feature of DARPA is active program managers overseeing, coordinating, and evaluating the R&D needed to meet the objectives.

The recent ARPA-H proposal appears to have its roots in work by advocacy groups, such as the Suzanne Wright Foundation, which is focused on pancreatic cancer and which championed an approach that it called “HARPA” to drive patient-focused innovation.  ARPA-H made it onto the presidential agenda, in part, through proposals channeled through the Day One project organized by the Science and Technology Leadership Council.  President Biden’s proposed FY2022 budget included $6.5 billion for the initiative, and Biden highlighted it as a way to “end cancer as we know it” in his first address to Congress in April 2021. In July, a House Appropriations subcommittee approved less than half of the request, $3 billion, for a new ARPA-H (as part of a $6.5 billion increase for NIH overall) — which still represents a substantial down payment and which might be increased as the bill proceeds to the Senate.

Proposals for a health ARPA are not new. While they have differed in details, they have embraced several aspects of DARPA, including strong program management to achieve specific clinical objectives. The current biomedical research system is dominated by the “crown jewel” of the federal bureaucracy, the $40 billion per year National Institutes of Health, which distributes funds through a system of two-tiered peer review rather than the ARPA program manager model. The NIH’s main approach—hypothesis-driven fundamental research, often involving molecular research at academic medical centers—is not designed to target urgent and specific health problems or to implement solutions.  NIH does fund clinical trials and health services research, but its core competence is funding investigator-initiated discovery and innovation rather than urgent and focused problem-solving. A separate concern is that the agency is conservative (favoring established lines of research over breakthroughs), slow (a yearlong grant review cycle, with resubmission common), and bureaucratic, traits that can be obstacles to advancing even important “basic” science.

With broad support from the Biden administration, NIH officials, advocates, and influential members of Congress, the new ARPA-H proposal has a good chance of becoming a reality. If so, several design choices need to be made along the way, and contentious issues resolved, through specifics in the legislation or during implementation.  Behind those choices lurk longstanding tensions in the political economy of biomedical research funding.

Most crucially, ARPA-H needs to figure out what it will actually do. Activities envisioned for ARPA-H include supporting costly R&D activities in the “valley of death” between basic research and industry work. It might become part of the NIH’s translational efforts: NIH Director Francis Collins has suggested ARPA-H could be like an “NCATS on steroids,” referring to the existing National Center for Advancing Translational Sciences at NIH. Or, it could be a way for the agency to more easily to do big projects like the human genome sequencing project, or to develop technology platforms. Collins and others have also viewed ARPA-H as a way to be bold and take risks, countering the inertia and conservatism of peer review. These laudable goals have high social returns, but they are quite disparate.  Can one new agency fill all of the gaps in the US biomedical research enterprise?

ARPA-H may be a stand-in to address several limits of the current NIH approach, which (together with the symbolism of DARPA as “something different that worked”) may explain its popularity. In practice, it will need a specific articulation of what it aims to do, and what types of activities are “ARPA-able” in biomedical research.  Even more important, the problem-oriented focus may force a reckoning with something the biomedical research community has historically avoided: define specific desired health outcomes, develop a transparent priority-setting process for ranking their importance, work backwards to identify the types of research and activities to attain them, and measure progress. If ARPA-H is created and housed at NIH, it could compel NIH to take the “health” outcomes implicit in its name more seriously than its historic propensity to interpret its mission as funding excellent research rather than improving health.

Whether ARPA-H should be at NIH at all is a second bone of contention. The patient advocacy groups who put it on the agenda, including the Suzanne Wright Foundation, have advocated for independence from NIH. The former head of DARPA has argued the ARPA approach is “a deliberate counterpoint” to existing models and “will not succeed if it is tightly integrated into its parent organization.” The concern is that one cannot cure what ails NIH from within: the threat of contagion from NIH bureaucracy and incremental approaches that could undermine the bold, risk-taking, project-oriented approach that is needed. Writing in Science recently, Collins, Presidential Science Advisor Eric Lander, and other administration officials made the counterargument for keeping ARPA-H within NIH to maintain complementarity between what the NIH already does and what ARPA-H might do. Having it elsewhere, they contend, could generate transaction costs, coordination problems, and duplication. More cynically, NIH may not want budget competition from an upstart separate agency that does things differently and is popular with patients and politicians.

To our knowledge, there is no strong evidence regarding whether the benefits of potential complementarities from having ARPA-H within NIH would outweigh the risks of cultural contagion. There is, however, an intriguing historical parallel. Fifty years ago, inspired by the Apollo missions to the moon and the World War II Manhattan Project, Mary Lasker and other disease advocates called for a “War on Cancer.” Then, as now, advocates called for a coordinated attack, driven by specific mandates, timelines, and goals. And then, as now, advocates called for a separate agency, expressing concern about “bureaucratic overlay” from NIH and that its approaches were “too cumbersome to manage mission-oriented R&D.” The NIH Director at the time, James Shannon, vigorously opposed separation and (like Collins and colleagues did for ARPA-H) emphasized complementarity, arguing “cancer research benefits from the interplay with other diseases of man.” The compromise solution, embodied in the National Cancer Act of 1971, involved keeping the National Cancer Institute (NCI) at NIH but giving the NCI its own bypass budget, creating a President’s Cancer Panel, and making the NCI director a presidential appointee. Similar approaches may be relevant now.  Even if ARPA-H is part of NIH (as seems to be the Administration’s preference), its director and program managers need considerable autonomy and separation from the NIH bureaucracy. A strong first director, from outside the NIH establishment, is crucial. Keeping the agency physically separate from the NIH campus (perhaps in Boston, San Diego, New York, Seattle, or elsewhere), just as DARPA was separated from the Pentagon, ought to help ensure autonomy and is being actively considered.

But there are also important differences between health and defense. Unlike DARPA, where there was one major buyer of weapon systems with a clear military mission helping set priorities (and, at least, early in the agency’s history, the major customer and implementer of technologies it developed), the exact health mission is not clear. In health care, many major technologies ultimately require costly regulatory approval, and development and implementation involve multiple heterogeneous users and agents.  While the federal government is the single largest purchaser of health care goods and services in the world, government does not control health care to the extent it does the military.  Moreover, hundreds of disease groups compete for resources.  Throughout NIH’s history, patient groups have actively lobbied Congress, demanding support for “their” condition, whether cancer, heart disease, diabetes, Parkinson’s, or Alzheimer’s, with different diseases having different days in the sun.  Hence NIH’s 27 institutes and strong, persistent bipartisan support in Congress.  Yet, this means that policy (and budgets) place a premium on political savvy, organization, and access to key politicians rather than disease burden or systematic needs assessment. While patient interest certainly generates support for biomedical R&D and keeps the ultimate purpose of biomedical research top of mind, competing groups also strain a fixed budget and complicate priority-setting. We have already started to see these dynamics in the discussion surrounding ARPA-H. Initially the proposal focused on cancer, Alzheimer’s, and diabetes, but the amyotrophic lateral sclerosis (ALS) community successfully persuaded Congress to add ALS to the list of specifically mentioned diseases, and there are now calls for a focus on rare diseases. There is tension between keeping patient group support (and benefitting from their input on what types of solutions are needed) and, at the same time, not being spread too thin — which may lead ARPA-H to focus on “general purpose” research that plausibly feeds into multiple areas. Here again, a well-thought-through and transparent priority-setting process, and a skillful independent director, will be crucial for long-term sustainability and effectiveness.

A final issue that will likely be contentious is contracting provisions. The administration proposal includes language allowing ARPA-H to use nontraditional mechanisms, including “Other Transaction Authority” (OTA), to contract with entities more flexibly than is possible through standard procurement contracts, including firms with needed technology and capabilities that are unable or unwilling to comply with federal procurement regulations. DARPA commonly used OTA, and BARDA (the Biomedical Advanced Research and Development Authority), another HHS office, used OTA to contract for a number of important COVID-19 technologies, including vaccines. This practice, however, is controversial, since OTA agreements can avoid public-interest safeguards on patent rights resulting from federally funded research. These safeguards (including Bayh-Dole “march-in” rights to make products available on reasonable terms) have not been used in practice by NIH, but there have been growing calls to do so in the context of high prices on drugs resulting from public funding. Though the issue is broader, debates surrounding taxpayer rights in publicly funded technologies have been most heated in the context of biomedical research. Widespread use of OTA by ARPA-H without public-interest safeguards may spark considerable controversy.

ARPA-H promises a major change in the way the US government funds biomedical research, providing types of funding and R&D management approaches that differ from those currently present in the biomedical innovation system. However, adopting the ARPA moniker alone will not be sufficient for success. The key challenge will be to incorporate the crucial elements of ARPA—including mission-oriented priority setting, risk-taking, measurement and evaluation, and active program management—into a biomedical research funding system that has historically resisted them. At the same time, the model will have to be tailored to reflect that the economics and political economy of health R&D are very different, and in ways more difficult, than defense.

About the Authors