Scholars and policymakers have recently begun to focus on the role federal agencies charged with health-related missions can play in the development of innovative health technologies and promotion of access to those technologies. Appreciating the expertise of agencies like the National Institutes of Health, Food and Drug Administration, and Centers for Medicare and Medicaid Services has expanded the range of tools contemplated by scholars who had previously focused largely on the United States Patent and Trademark Office.
Yet much of this attention has so far focused on the capacities of single agencies, acting alone. This Article expands the existing scholarly focus, considering not only the potential innovation-related goals to be achieved if each agency acts alone, but also the potential for collaboration across agencies. This Article advances that conversation, drawing not only on the health innovation policy literature but also on the growing administrative law literatures focusing on inter-agency and intra-agency coordination models.
This Article develops a taxonomy of potential modes of collaboration, demonstrating how these agencies, which lack formally shared regulatory authority over the innovation space, nonetheless may collaborate to promote incentives for innovation. These agencies do sometimes collaborate to advance goals which are common to the agencies. But they can do more. These agencies may complement each other, accomplishing together socially valuable ends that cannot be accomplished alone. By considering each agency’s core competency, this Article develops specific proposals for agencies to work together going forward.
This Article goes on to consider legal barriers to existing collaboration and potential procedural mechanisms for enhancing collaboration between these agencies to achieve health innovation policy goals. Ultimately, it argues that a combination of reforms, both internal to and external to the executive branch, might be most useful in advancing these aims.
Introduction
Few problems in healthcare policy have galvanized bipartisan attention in the way that drug pricing has recently. In the wake of scandals involving both sky-high prices for new drugs1 and enormous price increases on old drugs,2 the American people and politicians on both sides of the aisle agree that something must be done. A range of new policies has been proposed on the subject, policies that would give Medicare the ability to negotiate drug prices,3 speed generic drugs to market,4 or punish companies for price increases on existing drugs.5
Unfortunately, most of these proposals are unlikely to have a significant impact on the drug pricing problem.6 Proposals like these, targeted at the actions of a single administrative agency or institutional actor, address only one piece of a broader innovation policy puzzle. Proposals that would enable agencies to work together—to gather and share information, to set priorities for research, and to approve new products—have far greater potential to lower the costs of developing new health care technologies and to simultaneously improve patient access to such technologies. Solving the drug pricing problem and other health technology questions requires a new view of the relationship between administrative agencies and innovation incentives.
Rather than focusing on the innovation capacities of single agencies acting alone, this Article considers the potential of the National Institutes of Health (NIH), Food and Drug Administration (FDA), and Centers for Medicare and Medicaid Services (CMS) to promote both the development of new, innovative health technologies and access to those technologies in collaboration with each other. This Article draws not only on the health innovation policy literature but also on the growing administrative law literatures focusing on interagency and intra-agency coordination models. Yet in many ways this Article goes further, arguing that although simple coordination is itself desirable, active collaboration that goes beyond mere coordination can make possible a far broader range of innovative outcomes than can the actions of any single agency acting alone.
Existing scholarship has considered the expertise of these agencies and recast their missions as innovation-promoting ones, expanding the range of tools contemplated by scholars who had previously focused largely on the role of the Patent and Trademark Office (PTO).7 To be sure, the PTO and other agencies do take actions that influence the course of innovation.8 However, scholars of health innovation policy ought to (and increasingly do) focus on these health-related agencies—the NIH, FDA, and CMS—for at least four related reasons.
First, within the context of health technologies these three agencies loom relatively large when compared with the PTO or consumer protective agencies like the Federal Trade Commission (FTC). This is not true in all fields of technology. The PTO administers the patent system, a generally uniform innovation incentive,9 and so in technological fields without large federal bureaucracies, the PTO may well be of primary importance or the FTC’s enforcement authority may assume a greater role. However, in the context of health technologies, patents are but one baseline piece of a larger innovation ecosystem, and the other levers in the ecosystem are primarily governed by these three agencies.
Second, the one-size-fits-all patent system is a blunt tool for incentivizing innovation both generally and in health technologies in particular. Diagnostics,10 medical devices, and pharmaceuticals are not equally in need of a broad twenty-year term of patent protection,11 yet the patent system lacks the ability to award purposively longer or broader patents to innovators that need them to encourage innovation.12 By contrast, as explained in Part I, the innovation incentives administered by the NIH, FDA, and CMS are often finely specified and can award different amounts of grant money, different lengths of exclusivity, and different reimbursement amounts where warranted by the type of innovation and even by the particularities of the invention itself.
Third and relatedly, the NIH, FDA, and CMS themselves have more substantive authority to create and tailor these innovation incentives than does the PTO, in two ways. The PTO first lacks substantive rulemaking authority13 and the ability to tailor the patent system by area of technology.14 Meanwhile, the NIH, FDA, and CMS can all develop guidelines about which types of innovations will receive their support. But the PTO also lacks the institutional expertise to administer field-specific innovation-related programs of this type. Although the PTO has now begun to develop at least some related expertise with the creation of the Office of the Chief Economist,15 it cannot yet compete with agencies like the NIH, FDA, and CMS, which make both macro and micro decisions every day about where to allocate innovation resources.
Fourth and perhaps most importantly, these health-related agencies have the ability to promote innovation more broadly than the PTO is capable of. To be sure, “innovation” is a slippery and much-debated concept, but having a clear idea of agency capacity is helpful in defining this Article’s focus. Scholars have recognized that patents are most effective at encouraging innovation into highly excludable goods (like pharmaceuticals),16 and even then only where the relevant patient population has the ability to pay for the products.17 But patents are not particularly good at encouraging innovation where the social value of the technology cannot be captured by the patent-holder.18
This Article focuses on this class of innovations, those health technologies where private market signals are not likely to be reflective of social value. These administrative agencies have a key role to play where novel health technologies would primarily benefit the poor,19 or whose development can be expected to be particularly lengthy,20 or where the innovation is nonexcludable in some way (like a surgical method).21 In such cases, pharmaceutical or medical device companies can be expected to underinvest in technology relative to the societal burden of disease, because of the distortions created by the market and the patent system. These agencies can help restore that balance.
This Article proceeds in four Parts. Part I considers the ways in which the NIH, FDA, and CMS individually contribute to the progress of innovation in health technologies. Part I organizes and directs the existing literature and policy efforts to consider the different categories and types of innovative efforts these agencies have engaged in. Part II moves on to the ways in which these agencies contribute to innovation collaboratively, developing a taxonomy of the ways in which the NIH, FDA, and CMS collaborate at present. Part III takes the discussion from the descriptive to the theoretical, considering how these agencies, which lack formally shared regulatory authority over the innovation space, should expand the range of their collaborations to promote incentives for innovation. Part IV completes the analysis by considering potential procedural mechanisms for enhancing collaboration between these agencies for the purpose of achieving health innovation policy goals. Ultimately, it argues that a combination of reforms both internal and external to the executive branch might be most useful in achieving these goals.
I. Administrative Agencies as Individual Actors in the Innovation Ecosystem
Scholars of innovation policy are paying increasing attention to the ways in which agencies like the NIH, FDA, and CMS may all promote innovation-related goals, even as they were set up to serve a variety of other purposes.22 The process of innovation in health technologies is mediated through a complex federal bureaucracy, with these administrative agencies playing key roles in the development, approval, and reimbursement of health technologies of all types. This Part primarily considers the ways in which each of these three agencies individually serves to promote health innovation, emphasizing the different policy levers available to the agencies in question.
A. National Institutes of Health
In some ways, the NIH’s role in the innovation ecosystem is simple to explain: it pays for research. To be more precise, it pays for an enormous amount of the biomedical research performed in the United States, disbursing more than $30 billion in research funding each year.23 Yet this simple story must be disaggregated into its component parts. The vast majority of the funds awarded by the NIH function as what scholars sometimes call “push” incentives.24 These funds subsidize research inputs, paying for general research and specific development of health technologies before those products come to market.25 However, this research is of many different kinds, each of which is worth considering separately. Further, the NIH is beginning to explore the potential of awarding funds as a prize to a company after a product is developed, what scholars may refer to as a “pull” incentive.26 This Section will explore each of these aspects of the NIH’s portfolio.
The front-end, “push” research supported by the NIH is of at least three different types: basic, applied or translational, and regulatory. First, just over half of the NIH’s budget funds basic research, designed to improve our understanding of particular diseases, bodily processes, and human pathologies without focusing on the development of particular products.27 Before scientists can discover treatments for a disease (diabetes, for instance), they must understand how the disease functions in the human body. What has gone wrong in a person’s pancreas and islet cells to cause them to develop diabetes? How do those cells normally function? What are the symptoms of the disease, and how does it progress? Are some people more likely to develop the disease than others, and if so, why? Basic research questions like these are critical to the advancement of knowledge that only later leads to approved treatments.
Basic research has at least two special characteristics that make it particularly well-suited for support from the federal government. First, in many ways it is best understood as a public good. Basic research produces information, not products, and that information can and should be made available as widely as possible both to enable other scientists to make further basic science discoveries and to ensure that many companies with different ideas can rely on accurate scientific information in attempting to develop new therapies. Economists have long recognized that classic public goods like clean air and national security will be underfunded by private actors due to free-rider problems,28 and so as a result they are typically provided and supported by governments. So too with basic research of this type.
The second characteristic of basic biomedical research in particular is that it is extremely risky. Basic research may fail, scientists may discover something other than what they set out to find, and even successful research may take years to pay dividends. Scholars have argued that the long-term, uncertain nature of basic research makes it a good target for public funding.29 Essentially, the private market will view any potential dividends as both too uncertain and too far in the future to merit investment.30
The NIH devotes a little less than half of its budget to applied research,31 much of which seeks to translate basic research findings into medical practice. Although the line between basic and applied research is sometimes blurry, much of the NIH’s research related to specific health technologies clearly falls into the “applied” category. This includes research identifying potential new drugs by screening libraries of compounds, studying existing drugs for new uses, or supporting clinical trials of new drugs, devices, or diagnostics. Organizationally, the NIH has devoted an entire center to translational research through the National Center for Advancing Translational Sciences (NCATS).32
Much, though certainly not all, of NCATS’s work focuses on areas of research that are likely to be underfunded if private companies are the only source of investment, as was true in the basic research context. Consider NCATS’s New Therapeutic Uses Program as one example.33 Its goal is to engage in a practice that is often referred to as drug repurposing: to identify new uses for drugs that have already been approved or that have taken substantial steps toward being approved.34 Scholars like Professor Becky Eisenberg have previously explained that drug companies lack sufficient incentives to continue to study their approved drugs for new uses, because patents or FDA exclusivity periods covering those new uses are difficult to enforce, making investments in additional clinical trials difficult to recoup.35 In some ways, the goal here is similar to the goal in the basic research context: not necessarily to develop new drugs, but to develop new information about existing drugs. To the extent that such information will be underproduced by the private sector, this is another fruitful area for public funding.36
Other NCATS programs have similar characteristics. NCATS’s support of small businesses through the Small Business Innovation Research program may help democratize biomedical research, providing support to ideas with high social value but low commercial value that may be less attractive to traditional medical technology companies.37 Further, NCATS’s projects in the area of rare diseases help create infrastructure, such as repositories of patient data,38 and encourage collaboration through building networks of scientists working on these conditions.39 Here NCATS can also play a complementary role to the private sector by curating the relevant information and creating the conditions necessary for the process of drug discovery and commercialization.
Although it is commonly assumed that clinical trials are only overseen by private companies seeking to commercialize products, sometimes the NIH does fund and engage in clinical trials itself. Yet here again most of the work performed by the NIH is devoted to studies whose design will in some form be less attractive to the private sector. Maybe the technology under study is not a device or drug but a method of treatment, which is not so easily monetized by the private sector.40 Or perhaps the clinical trial is so lengthy that no pharmaceutical company is willing to engage in it, as by the time the trial ends, any patents on the drug in question will have expired.41 In either case, the NIH’s work complements the private sector’s activities.
The third category of NIH research, regulatory science, has received considerably less attention in the scholarly literature, but it may be the most important in terms of the NIH’s interactions with other agencies. Essentially, regulatory science seeks to study and develop new information, tools, and methods for use in the regulatory process, primarily the FDA regulatory process. The NIH’s existing regulatory science programs run all along the spectrum of regulation. For example, the NIH’s Tobacco Regulatory Science Program helps answer basic scientific questions about tobacco products that underlie much of the FDA’s regulation in this area.42 The Program helps answer questions about toxicity thresholds for different chemicals, the impact of product characteristics on decisions to start smoking, and the use of products like cigars, smokeless tobacco, and e-cigarettes.43 The results of these studies may both help set FDA priorities and may support specific rules that the FDA seeks to promulgate.
Other regulatory science initiatives study the process of regulation itself, seeking to make it more effective or efficient. The NIH has funded projects to speed the pre-clinical phase of the drug development process through their Tissue Chip for Drug Screening program, designed to more accurately model the functioning of human organs and reduce the time and expense of the pre-clinical phase.44 Even later in the drug approval process, the NIH has awarded grants for the study of innovative clinical trial designs,45 where the idea is to develop methods of conducting clinical trials that improve efficiency and patient safety. Here, too, this research is likely to be underprovided by the private sector, as any one company’s efforts in this space would redound to the benefit of its competitors.
Although the vast majority of NIH funding is disbursed as front-end push funding of the type described above, the NIH has displayed at least some interest in awarding funds after a product or method has been developed or at least conceived, as a prize or pull mechanism.46 The NIH recently awarded prizes for the conception and development of methods for analyzing single cells within a large population47 and for biomedical engineering advances,48 and it has announced a $20 million prize for the development of a rapid point-of-care diagnostic designed to reduce antibiotic prescribing.49 To be sure, the NIH’s expertise in prize administration is still inchoate. But the 2016 21st Century Cures Act50 directs the NIH to devote more resources toward so-called “Eureka prize” competitions,51 and so we can expect the NIH’s actions in this area to increase with time.
The NIH’s influence on research funding goes beyond the already large amount it disburses each year. The NIH in particular, and the federal government in general, have enormous soft power to set the agenda for the types of research that will be prioritized and engaged in by universities, foundations, and the private sector. Recently announced governmental programs like the Brain Research through Advancing Innovative Neurotechnologies (BRAIN) Initiative, the Precision Medicine Initiative, and the Cancer Moonshot provide helpful examples. These programs have marshaled substantial non-government funds in pursuit of their goals. For instance, non-governmental entities have contributed over $500 million to date to pursue the goals set forth by the BRAIN Initiative,52 bolstering the $300 million devoted by the government in 2016 to the project.53 Non-governmental projects related to the Precision Medicine Initiative may exceed the $200 million allotted by the government for that program, with institutions from hospital systems and universities to technology companies like Intel, IBM, and Microsoft all devoting resources to precision medicine projects.54
The NIH has decades of experience administering grants to promote innovation in the life sciences all along the spectrum of health technologies, from basic research to post-approval studies. Yet scholars (of science and law alike) do and should ask questions about how the NIH allocates funding among many competing priorities.55 The appropriate mix of funding between basic and applied research is an open question, as is how closely NIH funding should track disease burden.56 Problematically, these questions are often asked from a viewpoint that assumes the NIH is the sole funder of research in the United States.
Instead, discussions about the appropriate balance of NIH funding (both by research type and disease area) ought to begin by recognizing that the NIH, at least in some cases, complements research funded by other parties.57 In the area of basic research, the NIH may well be the primary funder by a large margin. But when considering translational research or regulatory science work, scholars ought to ask how the NIH might augment work being performed both in the private sector58 and, more generally, in the executive branch. The NIH’s work in regulatory science in particular may have a multiplicative quality. The development of new clinical trial models or of new methods for conducting preclinical research may reduce development costs not just for one but for a wide range of companies. Asking how the NIH might best use its limited funds to support projects with broad-based impact, as opposed to projects that would benefit one or a few private companies, is a key area of further policy discussion.
B. Food and Drug Administration
The FDA is too often criticized as an entity whose raison d’etre is to impede innovation by imposing onerous, expensive requirements on companies seeking to bring new products to market. This view is grossly oversimplified.59 In actuality, the FDA may be the most powerful innovation-promoting agency in the federal government when it comes to life sciences inventions, more important even than the PTO. The FDA promotes innovation in two types of ways: it forces information production through its role as regulator of drug and device approvals, and it encourages the development of particular types of socially valuable products through its administration of a series of finely specified innovation incentives.
First, the process of drug and device regulation overseen and administered by the FDA is indeed expensive and lengthy,60 although the cost and time vary widely by drug and risk level.61 However, inherent in the FDA’s authority to oversee this process is the ability to shape the kinds of information that companies produce through the drug approval process.62 More specifically, the role of the FDA as pharmaceutical regulator is to ensure pharmaceuticals and devices that companies seek to bring to market are safe and effective for their intended use. In fulfilling this role, the FDA requires companies to undertake clinical trials and produce information about their products. Professor Becky Eisenberg has reframed this role as “motivat[ing] drug sponsors to generate valuable information about their drugs.”63
This information—about which drugs and devices are useful for which patients at which doses or using which methods, or about whether any harmful side effects may arise, etc.—is useful, not only to the FDA and the individual company seeking FDA approval, but also far more broadly to scientists, policymakers, physicians, patients, and other companies. The company’s reward for producing information about the safety and efficacy of a particular drug or device is the ability to monetize the product itself. But the social value of that product also lies in the information produced through the clinical trial process, not just in the approved drug or device.64
The second set of ways in which the FDA promotes innovation is not inherent in its role as pharmaceutical regulator. However, Congress has nonetheless given the FDA the authority to administer a number of policy levers that provide incentives for companies seeking to develop pharmaceutical products.65 The primary goal of these mechanisms is to promote innovation incentives, unlike the information-producing role of the FDA, which has an incidental (but strong) innovation function but whose purpose is first and foremost the production of data to inform the agency’s safety and efficacy evaluations.66
As in the case of the NIH, the innovation policy levers administered by the FDA can be divided into push and pull mechanisms, some of which help decrease the costs of development and others of which provide rewards for successful approvals of particular types of products. In the push column, the FDA primarily administers a series of four expedited approval pathways.67 These programs—priority review, accelerated approval, fast-track designation, and breakthrough therapy designation—are designed to shorten the regulatory review process for companies seeking approval for drugs of various types.68 Shortening the review process enables companies to bring their products to market sooner, saving time and money and likely preserving more time remaining on their patent term. Some of the programs shorten the review process directly, as with priority review, which cuts the amount of time the FDA has to review evidence from clinical trials from ten months to six.69 Others shorten it indirectly, as with accelerated approval, which permits approval on the basis of a surrogate endpoint rather than a true clinical endpoint.70 Breakthrough therapy and fast-track designation shorten the review period in the process sense in that these designations entitle their grantees to frequent meetings with FDA officials on the progress of their drugs through the clinical trial process, ensuring there are no surprises in terms of the FDA’s demands come review time.71 Companies will often stack two or more of these expedited programs, combining their benefits.72
These programs are only available to companies producing drugs that meet two main criteria. First, they all require as a threshold matter that the drug in question be intended to treat a “serious condition.”73 Second, each program essentially requires the submission of evidence that the drug in question provides a significant improvement over and above existing therapies.74 The substance of this second criterion is worded differently across each program, yet they are phrased similarly enough that drugs will commonly meet several of them.75 For instance, the priority review program is triggered when a drug would provide a “significant improvement in safety or effectiveness” relative to current therapies,76 while accelerated approval is triggered upon a showing of “meaningful therapeutic benefit . . . over existing treatments . . . .”77
Most of the innovation policy levers administered by the FDA fall into the category of pull mechanisms. Two such levers in particular—exclusivity periods and priority review vouchers—are worth discussing in some detail, the first for its breadth and power and the second for its specificity and purposiveness.78 First, a number of statutes empower79 the FDA to award periods of regulatory exclusivity for approved drugs. In contrast to the uniform twenty-year term of patent protection,80 these periods vary in duration and in the set of drugs to which they apply. But they all serve the same function: to provide innovative drug manufacturers with sufficient incentives to carry new products through the extensive clinical trial process.
The Hatch-Waxman Act of 1984 and the Biologics Price Competition and Innovation Act, created as part of the Affordable Care Act in 2010, create data exclusivity periods for small-molecule and biologic drugs, respectively. Hatch-Waxman provides innovator small-molecule drug companies with five years of exclusivity, during which the FDA may not accept for filing an application for a generic product that uses the innovator company’s clinical trial data.81 The Biologics Act functions similarly but provides innovator biologic drug companies with twelve years of data exclusivity from the approval of the innovator product to the approval of the biosimilar.82 The appropriate length of the biologics exclusivity period was hotly contested,83 but in general there is agreement that exclusivity periods like these are useful in promoting long-term investment in drugs.
The Orphan Drug Act of 1983 is a closely related example. The Act provides special incentives to companies bringing to market drugs for FDA-designated orphan diseases, those affecting relatively small populations in the United States.84 Companies bringing such drugs through the approval process receive seven years of market exclusivity, during which the FDA may not approve a new or generic drug application for the same product and indication.85 Because the size of the patient market for orphan drugs is by definition small, the argument is that a period of exclusivity longer than Hatch-Waxman’s five-year period for new small-molecule drugs may be needed to assure pharmaceutical companies that they can recoup their investments.86
These exclusivity periods function in some ways like patents, and indeed they run concurrently for at least some period of time with any patent life remaining after a drug is approved.87 However, there are several key differences between the two. Exclusivity periods are automatically enforced by the FDA in its role as pharmaceutical gatekeeper, meaning that companies do not need to expend time and money searching for potential violators and then enforcing their patents against them, as they do in most technological fields. Additionally, exclusivity periods are nearly impossible to challenge in court, while patents are challenged by competitors and invalidated on a semi-regular basis. Finally, and perhaps most obviously, the exclusivity period is tied in scope to the marketed drug and its approved indication.88 Patents, by contrast, may be both broader than an individual drug (covering a class of compounds) and narrower than that same drug (covering only its method of production).89
The Priority Review Voucher (PRV) is a second pull mechanism administered by the FDA. A company receiving FDA approval for a treatment for any of a specified set of neglected tropical diseases90 receives a transferable voucher, which, when presented to the FDA, entitles its bearer to an expedited review process for a different product.91 The value of a voucher may be considerable, with two recent vouchers selling for $350 million and $245 million.92 Companies have used vouchers not only to come to market earlier, potentially having more patent-protected time on the market and saving costs, but also to leapfrog competitors and lock up market share earlier.93 Importantly, the voucher has been subject to several instances of gaming, in which companies merely shepherded existing drugs through the clinical trial process rather than developing entirely new compounds,94 but the central idea has economic merit.95
What is most interesting about the voucher is the list of diseases to which it applies and the FDA’s authority over that list of diseases. When first established, the list of conditions meriting a voucher overlapped largely but incompletely with the Word Health Organization’s list of neglected tropical diseases.96 Importantly, Congress foresaw the possibility that the FDA might wish to add diseases to the PRV list, and it authorized the FDA to designate by regulation “[a]ny other infectious disease for which there is no significant market in developed nations and that disproportionately affects poor and marginalized populations” as deserving of a voucher.97 The FDA has since exercised this authority, issuing in 2015 an order adding Chagas disease and neurocysticercosis to the list of designated tropical diseases such that manufacturers of drugs approved for these conditions may be awarded a voucher.98 In its order, the FDA both specified the factors it intends to consider in interpreting the terms of the statute and applied those factors to the two diseases at issue.99 The FDA’s experience fielding applications to add diseases to the PRV list and giving reasons in its adjudication thereof help demonstrate its expertise as an administrator of innovation incentives.100
Importantly, all of the above innovation incentives are largely limited to drug approvals. The vast majority of medical devices and diagnostics requiring FDA approval are not entitled to any exclusivity periods post-approval,101 and there is no analogous PRV program. Some medical devices may be entitled to expedited approval pathways similar to those existing in the pharmaceutical context, but the device programs are newer, smaller, and their impact remains unclear.102 This pharmaceutical exceptionalism is likely because pharmaceuticals are generally thought to be far more expensive, time-consuming, and risky to develop than almost any other product, including other health technologies. Policymakers have repeatedly been convinced of the need to create special pathways to benefit pharmaceutical companies for fear of decreased innovation.103
As a whole, many of the innovation incentives administered by the FDA are intended to bolster private incentives to invest in drugs either for which incentives would otherwise be too low or which may have a particularly high social value. The Orphan Drug Act and Priority Review Voucher program for neglected diseases are facially designed to support incentives for companies to engage in research on diseases affecting small, poor, or otherwise marginalized populations. The expedited approval pathways are designed to push companies toward products making truly significant treatment advances or meeting unmet medical needs, rather than the stereotypical “me-too” drugs which may add little, if anything, to the arsenal of drugs already available for a given condition.
C. Centers for Medicare and Medicaid Services
Unlike the NIH or FDA, CMS is rarely discussed in the literature as an innovation agency. Yet CMS has an important role to play in innovation policy, primarily through the enormous amount of money it brings to bear on the system. Specifically, CMS’s role as insurance company for over 100 million Americans through Medicare and Medicaid104 means that it makes countless decisions about which health care services and products to purchase and at what prices. Scholars studying CMS have therefore generally understood it as a tool for promoting access to existing medications. However, those reimbursement decisions also give CMS great power to influence ex ante the kinds of technologies that are developed.105 Further, CMS also has a secondary role to play in the innovation process by virtue of its possession of an enormous amount of health care information.
Most importantly, prescription drug insurance administered through CMS strongly resembles a prize system106 or pull mechanism rewarding innovator drug companies, much like those administered by the FDA. By law, Medicare and Medicaid are required to cover most (and for some classes of patients or diseases, all)107 FDA-approved drugs, meaning that companies know there is at least a partially guaranteed market for their drug. The size of the reward obtained is still uncertain, depending on things like the number of patients with a given condition, the distribution of those patients across the insurance system,108 and the number of drugs competing in a particular class.109 But the availability and size of the reimbursement pool is a key factor underlying pharmaceutical companies’ decisions to invest in drugs for different classes of diseases.
Medicare Part D is the clearest example of this phenomenon in the health insurance context. Although the broader Medicare program has existed since 1965, Medicare largely110 did not cover prescription drugs until 2006 when Medicare Part D went into effect.111 Medicare Part D provided a prescription drug benefit to Medicare enrollees, and as a result it both expanded the population of seniors with access to prescription drug coverage112 and increased the prices pharmaceutical companies could expect to recoup for many drugs sold to senior citizens who had previously been eligible only for Medicaid.113 The passage of Medicare Part D has been empirically associated with increased pharmaceutical investment in drug classes with higher consumption among the Medicare population.114 Medicare Part D is not the only example of this phenomenon, with other analyses examining the effects of individual coverage mandates or of population shifts.115
It would be a mistake to view the innovation-related effects of Medicare Part D merely as an accidental side benefit of its publicly stated goal to promote access to pharmaceuticals for Medicare enrollees. Public health insurance schemes exist to instantiate a series of policy choices about which populations deserve coverage and which technologies and services will be covered. Even if consumers value and are willing to pay for cosmetic drugs (such as injections that reduce the appearance of facial wrinkles or under-eye bags), governments may choose to manipulate the insurance system to disfavor these or other categories of drugs. Governments (including the United States) often make decisions to favor certain socially valuable categories as well, such as when the Affordable Care Act mandated that preventive technologies like routine vaccinations be covered without cost-sharing.116
A more purposeful example of CMS’s potential as an innovation agency comes through its implementation of the new technology add-on payment (NTAP) system in Medicare. Essentially, policymakers had become concerned that Medicare’s system of paying for hospital services did not sufficiently reward the development of new technologies and their incorporation into medical practice. As a result, Congress directed CMS to create a procedure for identifying new medical technologies and providing additional payments for their use.117
Although the NTAP statute is highly general, CMS has created finely specified procedures for implementing the program. By regulation, CMS has established criteria for determining which medical technologies are eligible for the add-on payments, created an annual application system for interested companies, and developed a formula for calculating the size of the payments.118 As the FDA has cultivated its expertise over time in administering expedited approval programs and the PRV system, the expertise CMS has developed in implementing the NTAP program could be brought to bear more generally to encourage innovation into needed medical technologies.
Beyond its role as the largest payor for health technologies in the country, CMS also possesses enormous amounts of information about the health needs of the American people in its capacity as their health insurer. Medicare and Medicaid have information about what the primary health care needs of their beneficiaries are, what diseases are imposing the largest financial drains on our system, and relatedly, where health innovation can provide the greatest benefit to the many Americans under CMS’s care.
Scholars have paid attention to the ways in which the NIH, FDA, and CMS individually play significant roles in promoting health innovation in the United States. They are in charge of administering different policy levers and working at different stages along the value chain of innovation, from funding basic research to shepherding products through the approval process, to paying for the results of investment and innovation. Yet scholars have paid comparatively less attention to the ways in which these agencies sometimes work together to promote health innovation.
This lack of focus on agency collaboration is puzzling particularly in light of the fact that each of these agencies is formally a sub-agency within the Department of Health and Human Services (HHS). As such, scholars and policymakers ought to consider not only the potential innovation-related goals to be achieved if each agency acts alone. They also ought to consider the potential for coordination across agencies under the auspices of HHS. This next Part takes up that idea.
II. Interagency Coordination and Collaboration in Practice
This Part aims first to construct a taxonomy of the ways that the NIH, FDA, and CMS work together at present to promote health innovation. Secondly, though, this Part argues that such coordination is often limited, non-public, and tenuous. Sometimes such relationships seem to be ad hoc. In other cases, public clashes exist to balance out the examples of coordination.
Before constructing the taxonomy, a word about terminology is in order. Much of the leading administrative law scholarship on interagency relationships speaks in terms of coordination,119 where the primary goal in situations involving agencies with interacting jurisdictional assignments (like the NIH, FDA, and CMS) is to minimize inconsistency.120 But it is worth distinguishing mere coordination from a relationship that rises further to the level of collaboration, in which agencies actively work together, exchanging information and resources, to achieve shared goals. To be sure, the literature itself (rather than just the terminology it applies) is concerned with both. Requirements for interagency consultation frequently fall into the coordination category while joint rulemakings are often closer to collaboration, and there are also particular instances of true collaboration.121 My focus going forward is on this category of true collaboration, although the personal and institutional relationships supporting relationships of mere coordination are often a key precondition for collaboration.
In constructing the taxonomy of forms of interagency collaboration, this Article is comparatively free from a problem that has affected most of the scholarship focusing on interagency relationships: the problem of agency non-disclosure. That is, agencies are often not required and often do not choose to disclose the many different types of relationships they have with other agencies.122 Some individual relationships can be discerned from particular statutes requiring consultation or public actions taken jointly,123 but in general, transparency is not the norm.
By contrast, the NIH and FDA are generally very forthcoming on the collaborations they have with other federal agencies. Since 2006, the NIH has been required by statute to publish an annual report identifying, categorizing, and analyzing every relationship it formally enters into with other administrative agencies.124 And although seemingly not required by law, the FDA publishes a list of all formal memoranda of understanding it has with other federal agencies.125 CMS is comparatively less forthcoming, but information about CMS’s collaborations with the NIH and FDA, in particular, may of course be backed out of those agencies’ lists. This is not to say that these lists are exhaustive—indeed, as I go on to argue in this Part, I suspect additional relationships are operating behind the scenes. Further, much of the publicly available information about these collaborations is quite sparse. However, these lists are a start and provide a window into the size and scope of interagency collaboration in this space.
In general, the NIH, FDA, and CMS collaborate in different combinations and ways to serve different functions. These functions can loosely be grouped into three main categories: information-sharing, research, and decision-making and product approval.126
A. Information-Sharing
More than two hundred collaborations devoted primarily to information-sharing purposes have proliferated among these agencies, and these types of collaborations are more prevalent than any other type of relationship between these agencies.127 These information-sharing initiatives sometimes seem to go by as many different names as there are collaborations—working groups, advisory committees, coordinating committees, leadership councils, etc.—but they all: (1) bring together representatives from more than one agency; (2) for the purpose of sharing information between them.
Although they have the same basic function, these groups differ from each other in a host of ways. Substantively, they span most of the health technology space, considering topics from pediatric drugs128 to nanotechnology129 to blood products.130 Structurally and procedurally, these groups differ along at least five key dimensions. First, they exist at different stages within the leadership hierarchy. Second, they exist at different levels of generality. Third, they were constituted through different types of procedural mechanisms. Fourth, they engage in varying levels of transparency about their activities. Fifth and finally, they use the information shared through the group for different purposes. That is, sometimes information-sharing serves to set priorities for further research. In other cases, information-sharing is used to revise existing agency processes. To illustrate these differences, consider the following two examples.
The NIH-FDA Joint Leadership Council brings together the heads of the FDA and NIH and senior leadership from both agencies131 to engage in broad-based consideration of the ways in which information-sharing can support the activities of each agency. Specifically, the Council aims to ensure both that “regulatory considerations form an integral component of biomedical research planning,” and that “the latest science is integrated into the regulatory review process.”132 Since the Council meets only at the request of the Chairs, each of whom is a busy agency head, it is difficult to say how productive the Council has been thus far. But it represents one possible information-sharing model. The Council exists at the highest levels of the leadership hierarchy, has a broad scope of interest, is not very transparent, and aims to use the information shared with the group to inform and improve both research priority-setting and regulatory processes.
A very different example is the Interagency Pain Research Coordinating Committee (IPRCC). The Committee’s creation was mandated by the Affordable Care Act, which not only instructed the Secretary of HHS to establish the committee “to coordinate all efforts within [HHS] that relate to pain research”133 but also finely specified the membership, meeting schedule, and duties of the committee.134 The Committee’s leadership is less senior than the Council’s and its focus is narrowed to pain research. And although within that focus the Committee aims to achieve different goals (including both prevention and treatment, for instance), it is primarily focused on bolstering research, rather than altering the regulatory environment for pain-related products.135 The Committee is also highly transparent about its activities. The Committee announces its public meetings in the Federal Register, streams the meetings online, and provides the public with access to the presentations and minutes.136 It also publicly outlines its strategic plan for pain research and solicits feedback on ways in which it can be improved.
B. Research
In some cases, the information-sharing conducted through various collaborations, as considered in Section II.A, will then be used by individual agencies in their sponsorship of different research or regulatory activities. But in other cases, the collaboration between the agencies is based around the research itself, rather than information underpinning that research. These research initiatives are diverse in their subject areas, focusing on everything from rheumatoid arthritis137 to robotics138 to women’s health.139
Although there was great structural and procedural diversity in the types of collaborations occurring in the information-sharing context, there is somewhat less variety in these research collaborations. Some of the homogeneity stems from the nature of research, which relies on the efforts of individual scientists asking specific questions, not agency heads promulgating vague directives. Therefore, these research initiatives are generally situated lower in the agency hierarchies and initiated with greater specificity than are some of the information-sharing collaborations. Further, because the vast majority of these collaborations exist between the NIH and the FDA (with CMS contributing only to a small handful), there is relative homogeneity in the type of research being conducted. Most of these collaborations involve regulatory science rather than active product development. To illustrate, consider two examples.
First, the Tobacco Regulatory Science Program (TRSP), as discussed in Section I.A, directs research into tobacco products that then supports the FDA’s regulatory activities. The TRSP was developed internally to HHS by the NIH and the FDA, rather than formally imposed externally by the President or Congress. Research supported by the TRSP can help the FDA set its regulatory priorities, decide whether to regulate a product at all, and decide how to regulate it. For instance, one of the TRSP’s areas of focus is e-cigarettes.140 Research into e-cigarettes, including their health risks, use by minors, and chemical characteristics, may support these aspects of the FDA’s decision. Although the FDA’s regulatory decisions may influence the direction that private industry takes in developing new tobacco products, the primary point of the TRSP is not to accelerate such innovation, but to choose how to regulate it.
Second, the BRAIN Initiative is an effort to “support[] the development and application of innovative technologies that can create a dynamic understanding of brain function.”141 Much like the Precision Medicine Initiative or the Cancer Moonshot, the BRAIN Initiative emerged from and is driven by the White House as one of its key priorities in biomedical innovation, rather than emerging from internal HHS discussions as the TRSP did.142 This allows the Initiative to bring together agencies within HHS (NIH and FDA) as well as other agencies outside of HHS, including the National Science Foundation and the Defense Advanced Projects Research Agency.143 The proximate goal of the BRAIN Initiative is not to develop treatments for brain-related conditions like Alzheimer’s disease or depression that exact an enormous toll on our health care system.144 The BRAIN Initiative seeks instead to develop tools and information about these diseases that can be used both by these companies and by regulatory agencies going forward.
C. Decision-Making and Product Approval
Far less common than either information-sharing collaborations or research initiatives are efforts to engage in collaborative decision-making and product approval. The NIH seemingly does not engage in collaborative activities that are directly related to the approval of individual products, but the FDA and CMS have recently developed two such collaborative programs in the context of medical devices and diagnostics. By statute, the FDA and CMS have shared jurisdiction over devices and diagnostics,145 in which the FDA is in charge of approval and CMS is in charge of reimbursement. The idea behind these collaborative programs is to harmonize, where possible, the often divergent standards employed by these agencies in making those decisions and to reduce uncertainty for companies bringing new products to market.146
Historically, medical device companies have had to navigate two separate regulatory systems: they must both obtain FDA approval and proceed through CMS’s national coverage determination process to secure Medicare reimbursement for their device. But the two agencies apply different legal standards to those determinations147 and are concerned with different aspects of the device, resulting in substantial uncertainty about whether the information generated in the FDA approval process is sufficient to support a coverage determination.148 Even where a company has produced sufficient information, the additional time required to go through the CMS coverage determination process after FDA approval is costly both for the company and for patients who may want to access the device in question.
As a result, the FDA and CMS have developed a collaborative decision-making program operating in the medical device space: parallel review. The Parallel Review pilot program, first created in 2011, allows product sponsors to request that CMS begin the coverage determination process while the product is still under review by the FDA. The idea is to collapse the two review timelines, at least partially, and permit product sponsors to anticipate and develop the data needed by both agencies. The program was formally made permanent in 2016,149 although in its five years of operation, just two devices have been approved through the program—Cologuard, a non-invasive colorectal cancer screening test,150 and a next generation tumor sequencing test from Foundation Medicine.151
The other collaborative effort between CMS and the FDA involves diagnostic tests. Although the FDA is typically viewed by the public as the chief regulator of medical technologies, CMS actually plays a far more prominent role in the area of diagnostic tests. Until 2014, the FDA essentially exercised no regulatory authority over a large class of diagnostics known as “laboratory developed tests” or LDTs, those which are “designed, manufactured and used within a single laboratory.”152 Instead, these tests were primarily regulated by CMS through the Clinical Laboratory Improvement Amendments of 1988 (CLIA),153 which requires laboratories and the diagnostics they perform to meet certain analytical validity benchmarks.154 The FDA always took the position that it had the authority to regulate LDTs, but it chose to exercise its enforcement discretion, meaning that diagnostic testing companies faced only a single federal regulator.155 In 2014, however, the FDA announced its intention to begin regulating LDTs. The FDA’s plan was to require companies to produce a different type of data about their diagnostics—data about their clinical validity.156
Although the FDA has decided not to finalize this particular regulatory framework, the FDA and CMS have nevertheless created a task force on LDT quality requirements.157 The goal is for the agencies to collaborate in modernizing their regulatory requirements, taking care to delineate the respective duties of each agency158 and to minimize duplicative efforts by the agencies.159 From the agencies’ public statements, it seems as if CMS’s role through CLIA would be altered slightly to focus more on the labs themselves, while the FDA would focus on determining each test’s analytical and clinical validity.
As will be apparent from the preceding descriptions, when agencies choose to collaborate, there are benefits and drawbacks to choosing different collaborative mechanisms that may make them more or less attractive as a means of promoting innovation into health technologies. Information-sharing initiatives are relatively low-cost and the agencies devote little if any formal resources to them other than the time involved to collect existing information and gather officials for meetings. Of course, these personnel resources may be more or less scarce. The “cost” to convene a meeting of the NIH-FDA Joint Leadership Council, bringing together not only the heads of two agencies but over a dozen senior officials, may be quite high. But the on-budget cost of these initiatives may still be zero, meaning that institutional commitment is the primary barrier to their establishment. However, information-sharing initiatives may be comparatively less likely to actually lead to the development of new technologies, since they are so early-stage.
With research initiatives, the primary question is the availability of financial resources for the program. Higher-level support within HHS, or even external support (such as from the President), may be needed to convince agencies to spend scarce resources on particular topics rather than others. When the collaboration is adequately resourced and staffed by committed agency members, it may be more likely to lead to productive research.
In some ways, product approval collaborations face both kinds of difficulties. They may not require direct additional expenditures of money as do research initiatives, and they therefore depend more on institutional commitment for their functioning. However, they do result in the agencies spending time, resources, and expertise on areas that they might not otherwise have to pursue. For instance, when CMS begins the national coverage determination process for a medical device that the FDA later declines to approve, that time is taken away from other potential CMS projects.160
Relevantly, nearly all of the FDA-CMS collaborations noted above emerged in the last few years of President Obama’s Administration, under the leadership of Commissioners Robert Califf (FDA) and Andy Slavitt (CMS). Commissioners Califf and Slavitt have repeatedly and publicly emphasized the importance of agency collaboration in these areas.161 Under their leadership, a CMS-FDA-NIH Trilateral Council was formed, to “provide[] a forum for agency leaders to . . . determine appropriate interfaces and collaborations . . . .”162 The many programs resulting from their collaboration are evidence of the potential power committed agency heads can wield in promoting innovation.
The NIH, FDA, and CMS clearly do engage in at least some amount of collaboration to promote innovation, broadly conceived. However, orderly, transparent collaboration may be the exception, not the rule. As noted above, these collaborations are often restricted to particular types of products or information. Further, not all collaboration is public, although the best explanation for an agency’s action may be that collaboration or at least coordination exists behind the scenes. Consider as an example a grant made by the NIH for the study of FDA regulation of microbiota transplants. This grant was made under the Human Microbiome Project, which is fully housed at the NIH and does not formally include the FDA.163 Yet it seems that the FDA has at least knowledge of, if not input into, the grant. The grant itself is intended to provide information to the FDA, and FDA observers attend meetings of the working group.164 But the FDA has taken several regulatory steps that would seem to contravene the existence of this grant.165
Perhaps more importantly, the agencies not only do not always collaborate, but in some cases they do not even agree about issues and may clash publicly over them. A recent clash between CMS and the FDA is perhaps the most obvious example of this in recent years. In 2012, the FDA and CMS began a very public battle over medical devices. The battle concerned Unique Device Identifiers (UDIs), particular identifiers given to medical devices. The goal of the UDI system is to improve patient safety by enhancing the ability of providers and the FDA to track and report adverse events, to more speedily identify problematic clusters of events, and to facilitate the process for recalling unsafe devices.
Congress enacted the UDI concept into law in the 2007 FDA Amendments Act,166 and by 2012 the FDA had developed plans to incorporate UDIs into data from medical claims to enable them to reap the benefits of the UDI system.167 There was just one problem: CMS objected to the inclusion of UDIs in claims data, on the grounds that it was both technically difficult and expensive.168 The agencies feuded publicly for years, and only in the summer of 2016 did they settle their dispute in the FDA’s favor.169
This feud should not have spilled over into public view. It is true that agency competition over substantive issues can serve a range of laudable goals.170 Vociferous advocacy for both sides of the issue can help the administration as a whole reach the best policy outcome. But when policy disagreements are aired publicly, it may harm the relevant actors in the regulated industry. Actors on the device side (medical device manufacturers and labelers) as well as on the provider side (payers and claims management organizations) need to trust the agencies setting forth the relevant policies. If payers know that CMS is not committed to implementing UDI policy, they may drag their feet or otherwise attempt to delay its implementation.
The public airing of the UDI feud may be attributed to at least two potential failures. First, although the FDA and CMS are separate agencies with separate leadership, they are both formally sub-agencies within HHS. The existence of a strong agency head should have prevented conflicts between sub-agencies from reaching the public. Of course, this conflict took place at a time of turnover in the office171 and while HHS was focused on implementing the complex provisions of the Affordable Care Act, many of which began to go into effect during this period.
Second, the FDA and CMS had failed to reach a compromise decision on the UDI question. The FDA and CMS are agencies concerned with the very same types of questions involving the very same types of technologies. How can we encourage scientists and industry to develop new medical technologies that improve human health? How can we bring these products through our approval process and how will we pay for them? Their positioning as sub-agencies within HHS should have highlighted their overlapping missions, not pitted them against each other.
The reasons why CMS eventually came to agree with the FDA are not publicly known, but there are several potential explanations. CMS may have been swayed by recent scandals involving medical device adverse events, such as Olympus Corporation’s duodenoscope that caused an outbreak of antibiotic-resistant bacteria, infecting hundreds of patients and possibly killing several.172 CMS may have had an eye toward the potential cost savings from UDIs as it considers its push for value-based payments, which often penalize physicians and hospital systems for preventable adverse events. The collaborative relationship between Commissioners Califf and Slavitt may have catalyzed the agreement.173 Alternatively, the Secretary of Health and Human Services may have mediated the dispute between them.174 A strong HHS Secretary may have the power to defuse intra-agency battles,175 achieving coordination if not collaboration.
III. Toward a Purposive System of Interagency Collaboration for Innovation
Having considered in Part II the ways in which collaboration is happening between these agencies, this Part moves to consider the ways in which collaboration can and should occur but is not happening. The goal is to view the NIH, FDA, and CMS as agencies with particular core competencies and expertise and envision them in complementary situations. In doing so, this Part theorizes potential collaborations to fill the negative spaces in the existing collaborative landscape. Further, it argues that by working together, these agencies can achieve two unique types of goals: First, these agencies can accomplish goals together that they cannot accomplish apart. And second, by working together these agencies can accomplish goals synergistically, in the sense that their combined contributions are greater than the sum of any particular part.
The essential question asked in this Part is simple: to the extent that these agencies working together can more effectively promote innovation incentives than agencies acting alone, when should these agencies work together and in what capacities? I propose four categories which represent different modes of collaborations between these agencies. Importantly, the collaborations described in Part II fall largely into the second and third category, suggesting that although there is surely work to be done in those areas, large swathes of collaboration remain nearly unexplored.
A. Enhancing NIH Priority-Setting for Basic Research with CMS Data
The first category of collaboration brings together the NIH and CMS to engage in information-sharing and priority-setting activities. The central idea here is that when the NIH makes decisions about how to allocate funding between or within institutes, it ought to do so on the basis of data. To be sure, reasonable people can disagree about what types of data will be most important,176 how much weight the data should have in funding decisions, and about how to trade off different pieces of information against each other, a topic to which I will soon turn. But in general, the more information the NIH possesses, the better and more informed its allocation of resources will be.
Helpfully, CMS is in possession of an enormous amount of data that is relevant to the NIH in this area. In its role as health insurer to over 100 million Americans,177 CMS and its contractors have information about disease burdens, about what types of drugs and interventions exist for a particular disease, and therefore by extension about which disease areas are currently underserved by medical technologies. But there is also information that CMS may want but does not have. For instance, where there is more than one intervention available for a given disease, CMS may be uncertain about the relative value of those interventions.
Information-sharing between CMS and the NIH can better enable the agencies to accomplish two primary goals. First, the NIH may improve its ability to set its funding priorities if it has an increased understanding of the unmet disease burden faced by the many Americans on public health insurance. Second, CMS may be able to make more informed decisions about relative reimbursement rates if it has information from the NIH on cost-effectiveness and value.
Each of these potential goals may be illustrated with a single example: mental health conditions. Mental health and other neuropsychiatric disorders are now responsible for the loss of more disability-adjusted life years worldwide than any other set of conditions.178 Mental health disorders are responsible for at least $300 billion in costs annually in the United States alone, when the cost of direct health care expenditures is added to the (far larger) related lost income and disability expenses.179 Much of these costs are traceable to major depression, which affects approximately fifteen million Americans.180
Fortunately, the NIH already recognizes the high disease burden due to mental health conditions and devotes significant resources towards these conditions, devoting funding roughly in proportion to disease burden181 and devoting institutional resources in the form of the National Institute for Mental Health. Yet these investments have yet to lead to real, effective treatments for many patients with these conditions.182 The Director of the National Institute of Mental Health has argued that the new generation of antidepressants is no more effective than the medications available in the 1980s.183 We simply lack an understanding of neurobiology that can be translated into effective treatments,184 and these scientific difficulties have even led many large pharmaceutical companies to shutter their neuroscience divisions entirely.185
As discussed both in Parts I and II, the White House’s leadership of the BRAIN Initiative is an attempt to solve some of these problems, focusing on increasing our understanding of neurobiology and the development of tools that can eventually be used to create and evaluate new treatments. And without the White House’s leadership, it is difficult to say whether NIH would have reoriented its funding around these foci. But CMS is not involved in the BRAIN Initiative, at least not publicly.186 Instead, the FDA plays a prominent role in the Initiative.187 This may reflect the Administration’s view that pharmaceuticals are likely to be a fruitful avenue for treating mental health conditions—and it is surely right about that. But this is also a narrow view of medical interventions, one that CMS’s involvement would broaden.
To be more specific, there are interventions, such as pharmaceuticals and medical devices, where the private sector has historically played the dominant role in product development, manufacturing, and sales. Scholars typically argue that the incentive system we have developed around medical technologies—patents, exclusivity periods, and the like—is necessary to encourage private companies to invest in these types of products, because these exclusive rights permit companies to prevent others from making and selling their inventions.188 The system may not always be sufficient, though, and where the basic science is as uncertain as it is in the area of mental health conditions, additional early-stage research may well be warranted.
However, patents and other exclusive rights have a tendency to “predictably and systematically distort private investment decisions . . . by overstating the value of highly excludable information goods and understating the value of highly nonexcludable ones.”189 And there are a range of nonexcludable technologies with the potential to mitigate the symptoms of mental health conditions, technologies in which the private sector has displayed little interest to date. Consider talk therapy, which may be prescribed for the treatment of a broad range of mental health conditions. Yet there are few rigorous clinical trials examining the practice, including the ways in which it might be optimized for particular maladies.190 Even further along the continuum of nonexcludability, consider that studies have demonstrated the efficacy of exercise for the treatment of moderate depression, finding it as or more effective than existing pharmacological interventions.191
Not only is private industry largely uninterested in pursuing such nonexcludable technologies, but the FDA does not “see” interventions like these, as they do not fit within its purview of regulating and approving medical products. The FDA by law does not regulate the practice of medicine itself.192 But in some ways, CMS does. CMS is formally agnostic as between whether a decrease in disease burden is achieved through a drug or through a medical treatment,193 and CMS involvement in the BRAIN Initiative specifically or NIH decision-making more broadly may be valuable in raising the profile of interventions like these within the funding process.194
A CMS-NIH partnership may be able to go further in the area of pricing, enabling CMS to make more informed decisions about relative reimbursement rates for interventions based on cost-effectiveness and value. New pharmaceuticals are often evaluated against a placebo or only a single other treatment, and scholars have long recognized the dearth of quality research on the comparative effectiveness of different kinds of drugs.195 But CMS may not only be interested in comparing drugs to each other. CMS may also want to compare drugs to non-pharmaceutical interventions or explore combinations of pharmaceutical and non-pharmaceutical care. Just as there is little private investment in non-pharmaceutical interventions now, there is similarly little to no investment in the comparison of pharmaceutical and non-pharmaceutical interventions. Yet this information would benefit CMS as it sets reimbursement rates and makes coverage decisions.196
CMS may also work with NIH to ensure it obtains preferential rates for drugs or other technologies developed with the assistance of federal funds. Scholars and policymakers have often asked why the public “pay[s] twice” for some drugs, first by subsidizing their development and then again by paying monopoly prices for the resulting product.197 More recently, at least two Senators have introduced bills which would require companies to offer reasonable prices on technologies developed with taxpayer money more generally, not only to public payers.198 But as CMS has an interest not only in the development of new technologies for those Americans it insures but also the procurement of those technologies at prices that make them widely available, it would be fruitful to see CMS and NIH work together to ensure reasonable pricing terms at least for public payers.
B. Improving FDA Regulatory Processes Through NIH Involvement
The second set of collaborations would bring together the NIH and the FDA for two purposes that are similar to those expressed in the previous subpart. Specifically, the two agencies should first engage in information-sharing and priority-setting activities. The FDA is an expert on its regulatory process, but it also knows the ways in which that process could be improved. Sharing information with the NIH about difficulties with the current system and asking questions about the ways in which it can be improved would aid the NIH’s funding allocations in the area of regulatory science and would redound to the benefit not only of the FDA and the regulated industry but also the public, if medical products can be approved more efficiently and with greater accuracy. Second and relatedly, the two agencies should focus their efforts in terms of translational and regulatory science research on the particular areas of research that would help improve the regulatory process or at least gather information about that process.
These will not be new suggestions, given Part II of this Article. Unlike NIH and CMS, NIH and the FDA are already actively collaborating with each other across several dimensions. The NIH-FDA Joint Leadership Council creates space for high-level NIH and FDA officials to share information and communicate about joint goals, while the Interagency Pain Research Coordinating Committee (IPRCC) provided just one example of the ways in which these agencies can share information in a more concrete way.199 Further, the agencies are already engaged in different regulatory science research activities. The TRSP and the BRAIN Initiative are examples of the different ways in which research can support the FDA’s regulatory processes—in the first case, by supplying information that enables the agency to promulgate regulations and set priorities, and in the second case, by developing tools to be used in the process itself.200
However, there may still be two main areas of improvement for these NIH-FDA collaborations: purposiveness and transparency. First, the set of NIH-FDA collaborations that currently exists is not clearly driven by a high-level, systematic view of the areas in which these agencies ought to collaborate. Instead, NIH-FDA collaborations often develop in an ad hoc, externally-driven way. The IPRCC and BRAIN Initiative are examples of this—recall that the IPRCC was created and given specific charges by Congress and that the BRAIN Initiative is largely driven by the White House, rather than HHS.201 Although the NIH and the FDA may agree that these are worthy areas for them to invest time and resources, if left to their own devices they may have focused on other areas of collaboration.202
Instead, a higher-level strategy driven by HHS or at least by the heads of the NIH and the FDA would be more effective in achieving innovation policy goals. Conversations among leadership in which these agencies prioritized areas of information-sharing and research are essential to creating the most valuable data-driven, purposive collaborations for innovation in health technologies. Importantly, such conversations may already be taking place. The NIH-FDA Joint Leadership Council, bringing together the heads of the FDA and NIH and senior leadership from both agencies,203 seems perfectly positioned to have exactly these kinds of conversations. However, the main problem with the Leadership Council is its total opacity.
In addition to improving their purposiveness, the current set of NIH-FDA collaborations can be improved by increasing their transparency and public accountability. It is not a matter of public record when the Leadership Council meets—or even if it has ever met—and if it does meet, what the Council has discussed. Nor are there reports, annual or otherwise, detailing the Council’s areas of focus, particular projects, or research investments. Most of the publicly listed collaborations are closer to the Leadership Council’s level of transparency than to the IPRCC’s, and as argued in Part II, there are at least some instances of coordination which are not even publicly listed.204
There are good reasons to make these and other categories of intra-HHS collaborations more transparent, although these reasons may suggest different levels of transparency. First, transparency both about goals and about methodologies—such as whether these collaborations prioritize certain types of diseases over others, and if so, why—is important to ensure public accountability. These agencies already issue guidance documents and rulemakings for public comment for similar reasons. Issuing yearly reports or even documents listing the areas the collaborations hope to focus on going forward, as the FDA does each year,205 would be a step in the right direction.
Second and possibly more importantly, transparency about areas of collaborative focus helps academic scientists and companies plan for the future. The primary point of these collaborations is to improve both regulatory priority-setting and the regulatory process itself, and many of these collaborations may lead to disease or field-specific process improvements. Scientists and companies who have previously avoided particular disease areas or are considering exiting them due to regulatory barriers may reevaluate those decisions on the basis of new information about agency priorities and investment in these areas. In some ways, this is an exercise of the government’s agenda-setting power.
C. Reducing Uncertainty for Product Sponsors Through FDA-CMS Collaboration
A third set of collaborations unites the FDA and CMS to engage in activities that simplify and speed the two regulatory hurdles that companies aiming to produce innovative health technologies must typically surmount: approval and reimbursement. Companies approach each of these regulatory processes at different times in a product’s lifecycle and with different amounts and types of information about the product in question. But the separation of the regulatory hurdles adds time, cost, and uncertainty to the innovation process.206 The FDA and CMS have recognized the burden this additional time and uncertainty may place on innovative medical technology companies and have begun some collaborations in this area.207 Those collaborations themselves leave something to be desired.208 Even more problematic, though, is that the existing collaborations leave open broad areas in which the FDA and CMS could fruitfully collaborate but have not yet chosen to.
As detailed in Part II, in the medical device space, the FDA and CMS have entered into a set of collaborations designed to decrease the burden on companies from surmounting two separate regulatory hurdles with distinct evidentiary requirements.209 Specifically, their Parallel Review pilot program for medical devices aims to collapse the approval and reimbursement processes, providing earlier information to companies about what types of evidence will be required in both situations.210 And although their task force on LDT quality requirements is focused only on the approval part of the process, it seeks to harmonize and clarify the duties of both agencies in the review and approval of diagnostic tests, minimizing administrative and regulatory burdens for diagnostic companies.211
The Parallel Review program was finalized in late 2016,212 even though just two products have successfully been approved in the program’s five years of operation. It is difficult to say why the program has not yet been more successful. Perhaps industry was less interested in parallel review than they claimed to be publicly.213 Perhaps the program was inefficient, given the cost and rarity of national coverage determinations. More pessimistically, perhaps the agencies chose not to invest enough in the program to make it worthwhile. Or perhaps the FDA-CMS battle over the Unique Device Identifier question bled more deeply into the device sides of the agencies, making it difficult to work together. The lack of transparency over the program makes it impossible to learn from this experience. I reserve judgment on the LDT task force, as it is much newer. But it may be subject to these same concerns.
The FDA and CMS have at least attempted collaborations in the medical device and diagnostic areas. No such collaboration has taken place in the pharmaceutical context, at least not publicly. However, collaboration between these agencies in the drug context may have greater societal benefits than collaboration in the device or diagnostic space. The high cost of developing drugs and related failure rates have been widely publicized, and any collaboration with the potential to reduce those costs might be extremely socially valuable.
To be clear, there are plausible reasons why no comprehensive collaboration has yet been developed in the drug approval area. First, CMS is often statutorily required to cover particular FDA-approved drugs.214 As such, if companies view the CMS coverage determination as largely pro forma, with the primary uncertainty lying only in the FDA context, they may be even less interested in parallel review here. Second, because most drugs fail at some point in the regulatory process,215 it may not be an efficient use of CMS’s time to involve itself in the clinical trials process.
Yet there may be particular areas where consultations between the FDA and CMS might be useful. Three examples are illustrative. First, recall that the FDA administers a set of four expedited approval pathways, whose goal is to shorten the regulatory review process for companies seeking approval for drugs.216 The criteria for acceptance into these expedited pathways involves medical determinations that can be bolstered with evidence from CMS. Most importantly, each pathway requires the submission of evidence that the drug in question provides a significant improvement over and above existing therapies.217 Although each program words this requirement slightly differently,218 CMS is well-positioned to help the FDA make each of these determinations.
Second, although CMS is often statutorily required to cover FDA-approved drugs, in some cases CMS has a choice of which drugs to cover.219 In addition, even where CMS must cover a particular drug, that drug may receive a disfavored placement on a formulary and therefore be accessible to fewer patients.220 Companies who foresee that their drugs may fall into these categories may desire CMS’s input into the FDA regulatory process. CMS may provide feedback on whether the evidence required by the FDA for approval is sufficient to ensure the company a preferred formulary placement, or on what evidence would be required to demonstrate a drug’s cost-effectiveness relative to other products.
Third, CMS could assist the FDA in carrying out post-market reviews of approved drugs. Increasingly, as discussed in Part I, drugs approved by the FDA are using accelerated approval pathways that result in approvals based on surrogate endpoints.221 These drugs must be monitored post-approval to be sure that they are eventually proven safe and effective for the true endpoint they aim to cure—to be sure that the cancer patient whose tumors have shrunk will actually live longer, or the patient with heart disease whose cholesterol is lowered actually has a reduced risk of heart attacks. CMS, with its possession of and access to vast databases of patient data, can assist both industry and the FDA in making these determinations more efficiently.
D. Building Capacity and Lowering Costs with Tripartite Agreements
The final set of collaborations would unite all three agencies—the NIH, FDA, and CMS—to pursue a more advanced, more involved set of innovation goals. Together, these agencies have the capacity essentially to serve as a product development partnership, in which they set priorities in underserved areas of medicine, conduct research or make grants in those areas, and shepherd products through the FDA approval process. The end goal is to approve products which are then marketed at relatively low prices222 for unmet medical needs.
Collaboration at this level is not merely aspirational. The agencies are already engaged in partnerships that embody these ideals. One representative example is the Clinical Islet Transplantation Consortium,223 a network of clinical centers started in 2004 devoted to improving the safety and success of transplanting islet cells, the pancreatic cells that produce insulin, for those with Type I diabetes. Type I diabetes is typically diagnosed in children and young adults, and affects more than one million Americans.224 Although Type I diabetes is far less prevalent than Type II diabetes, which is typically diagnosed in adulthood and affects more than thirty million Americans,225 it still significantly burdens our health care system. The Juvenile Diabetes Research Foundation estimates that Type I-associated health care costs in the United States may be as high as fourteen billion dollars annually,226 and for the approximately 200,000 children and adolescents227 who cannot lead normal lives and who must either inject themselves with insulin several times each day or wear an insulin pump connected to their body, the psychological burden may be quite large as well.
Pharmaceutical companies, enticed by the large markets and daily needs of the diabetes population, have developed a range of pharmaceutical therapies for both Type I and Type II diabetes patients.228 But these therapies are often costly. Indeed, both scholars229 and the popular press230 have recently noted the skyrocketing cost of insulin. Further, adherence may be difficult to sustain, especially for children. And these therapies are just that—therapies, requiring patients to maintain these regimens as long as they live. Such therapies are intensely profitable for pharmaceutical companies but impose financial and other burdens on patients. Physicians and patients would prefer a cure.231
Islet cell transplantation may be such a cure. Type I diabetes patients’ bodies turn on and destroy their own islet cells. Successful islet cell transplants could enable these patients to avoid multiple insulin injections every day.232 Islet cell transplantation would not be a worry-free cure—at least for the foreseeable future, patients would still need to take immunosuppressive drugs as with other transplantation procedures—but it could be an improvement over current therapies.
Yet private industry is unlikely to devote significant resources to developing safe, effective methods for islet cell transplantation. Recall the discussion in Section I.A about pharmaceutical incentives to invest in new drugs rather than investing in the development of information about new drugs.233 Pharmaceuticals are excludable goods in the sense that they can be manufactured and sold to particular patients. Surgical methods are largely non-excludable in the sense that it is difficult to prevent the dissemination of information about how to perform a given surgical procedure once that information has been developed.234 As a result, it is difficult for private companies to monetize such information and they will be reticent to invest in its development.235 This is a classic situation in which public investment may be needed to drive the development of new medical procedures, and the NIH, CMS, and the FDA have come together for that purpose.
This example can be generalized to three broad areas in which these agencies might collaborate in a way that complements existing private sector investment for the purpose of developing products with high social value. First, the agencies might focus on developing diagnostics and drugs for conditions primarily affecting poor Americans, where CMS and state governments bear the primary burden of care but where pharmaceutical companies underinvest relative to the disease burden. Some of this work might focus on tweaking existing drugs or drug protocols with an eye toward improving adherence, for instance, but some of it might focus on the development of entirely new therapies.
Second, the agencies ought to focus on products whose development can be expected to be particularly lengthy. They might focus on preventive interventions for diseases like cancer or Alzheimer’s, where any study will need to be extremely large and possibly last for decades. Private actors are typically thought to be disincentivized from funding and performing such studies because any patent rights they may have in the product at issue may expire before the product comes to market, impairing their ability to recoup their investment in the technology.236 Yet these interventions may be highly socially valuable and deserving of research dollars.
Third and relatedly, the agencies ought to focus on investments in nonexcludable technologies, to the extent that they would implicate the FDA in some fashion. I have already argued earlier that CMS and the NIH should collaborate to direct research toward nonexcludable technologies that may prove valuable, particularly to CMS as an insurer. The Clinical Islet Transplantation Consortium described above is one such technology, but because it involves the transplantation of human tissue it necessarily implicates the FDA’s jurisdiction.
I am relatively agnostic about the methods the agencies use to form and carry out these collaborations. That is, one option would be for the government to essentially become a pharmaceutical manufacturer, developing the capacity to produce and sell any technology it develops in the above categories. Another option would be for the government to form partnerships with generic manufacturers and pay them to produce these therapies still at relatively low cost.237 Which route to pursue depends on a range of empirical factors including the economics of the situation as well as political priors about the role of government in general.
IV. Facilitating Interagency Collaboration
Given that there are tremendous unexplored opportunities for collaboration among these agencies, why have they not taken full advantage of these possibilities? This Part first considers existing barriers to collaboration. Are they practical, legal, or both? In light of such obstacles, how should we determine when collaboration is worthwhile? This Part then turns to procedural facilitation of such collaboration, presenting a menu of possible procedural approaches at different levels of hierarchy both within the executive branch and outside of the executive branch.
A. Legal and Practical Barriers to Collaboration
There are at least three potential barriers to interagency collaboration for the promotion of health innovation. The first is legal, the second practical, and the third a combination of the two. First, a number of laws constrain agency jurisdictions or agency information-sharing in a way that may impede collaboration. For instance, the FDA is not permitted to consider costs in the drug approval process,238 which may restrict its ability to engage in collaborations with that explicit purpose.239 Perhaps more problematically, the Trade Secrets Act240 and Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule241 may limit the information agencies are able to share with each other about potential products under development or about American health care needs.
A second set of barriers to interagency collaboration is practical. Interagency collaboration is difficult. It requires support from senior officials within the agency, scarce personnel resources, and (often) even scarcer financial resources. As considered previously, it may not be “worth” CMS’s time to help the FDA adjudicate applications for expedited approval as most drugs will fail before reaching the market. Relatedly, agencies may need to invest great time and energy in testing and improving collaborative systems as with the CMS-FDA Parallel Review program, which existed in pilot form for five years before it was made permanent.
Collaboration may be difficult for agencies who want to retain decision-making authority in their core competencies. The NIH understandably may not want to cede priority-setting authority to the FDA and CMS, especially when it faces so many continuing disbursement obligations. Political considerations may come into play here as well. To the extent that collaborations not only require buy-in from the highest levels of the agency, but require sustained buy-in over a period of time that may span administrations, new political appointees eager to make their mark on an agency may choose to jettison existing programs.
More problematically, collaboration may even require agencies to make trade-offs against their own core interests. Recall the fight between CMS and the FDA over Unique Device Identifiers, considered above. Although the Identifiers might help the FDA improve patient safety, in the short run they would only add costs to CMS, which must now alter insurance claim forms to include that field. The FDA may be forced to make similar trade-offs. Recall the various expedited approval pathways administered by the FDA, considered above. Some methods of shortening the FDA’s review period may result in the production of less information about the products in question, working against the FDA’s public health mission.
A third set of potential barriers may result from the interaction of legal and practical considerations. Note that the legal barriers raised above do not explicitly prevent the agencies from working together. Indeed, Congress’s requirement that the NIH both engage in and report on its efforts242 suggests that Congress would prefer more, rather than less, collaboration. Instead, the legal barriers merely make it more difficult procedurally for the agencies to work together. Memoranda of understanding signed between the various agencies recognize that some information, including information subject to trade secret protection or HIPAA, must be “protected from unauthorized disclosure,”243 and that “[s]afeguards are needed”244 for this purpose.
Collaboration is not an unalloyed good to be pursued at all costs. Whether any particular collaboration is desirable depends on an assessment of its potential benefits as well as of its potential costs to the system. We may well decide that an agency head who does not want to cede authority to other agencies should not be able to derail a meaningful collaboration. At the same time, we ought to tread carefully before requiring agencies to do work that is likely to be wasted when a drug fails in the clinical trial process. But it is surely not the case that we have arrived at the proper level of collaboration through trial and error. More can and should be done.
B. Procedural Approaches to Encourage Collaboration
To the extent that significant practical barriers must be overcome in order to encourage interagency collaboration, I now turn to policy options that would break down those barriers. These options primarily differ in terms of their location within the government and the range of actions they have the potential to address.
1. Procedural Approaches Internal to HHS
On its own, HHS could select among a number of procedural options that would promote substantive interagency collaborations. These options exist at different levels of formality and require different levels of involvement from HHS. However, they also have the potential to achieve different levels of success, as a result. Perhaps most importantly, though, options like these are not mutually exclusive—HHS might decide to select more than one such option.
First, the HHS Secretary might require each agency to set priorities for interagency collaboration and to report on those priorities regularly, perhaps annually. Each agency would need to identify particular areas in which it hopes to make progress and with whom it will work on those goals. This idea combines elements of other existing reporting obligations. As considered above, the NIH is required by statute to report each year on the collaborations it engaged in the previous year.245 This analysis is retrospective, but its existence does force the NIH to report out certain statistics about its collaborations. The FDA, by contrast, has developed expertise in setting strategic priorities for years in the future and publicly reporting on those at different levels of generality, from publishing broad five-year plans246 to making public lists of specific rules they hope to promulgate and benchmarks they hope to meet in the next year.247
Second, the Secretary might appoint a staff member—perhaps under the position of Chief Innovation Officer—whose sole responsibility is to identify potential avenues of collaboration within sub-agencies of HHS and is given the authority to summon representatives from those different agencies to meet and consider those issues. This idea draws inspiration from a proposal by Professors Arti Rai and Stuart Benjamin, who suggest the creation of a similar entity within the executive branch.248 My proposal does not necessarily conflict substantively249 with the Rai and Benjamin proposal.250 Their proposal, if implemented, might well contemplate the appointment of individual officers with subject-matter expertise in particular areas. If so, my proposal might be viewed as fleshing out the duties of the health innovation officer. More generally, it is important to reaffirm the importance of developing deep subject matter expertise in health innovation in a way that a more general member of the executive branch might not inherently possess.
A third, more reactive option would be for the Secretary to simply monitor and manage or adjudicate disagreements between sub-agencies as they arise. Consider the public dispute between CMS and the FDA over the use of Unique Device Identifiers, as explored in Part II as one such example. This dispute is not only unproductive and time-consuming for the regulated agencies, but it is also problematic for the regulated industry. Rather than allowing these disputes to fester for years, the Secretary could take a more active, top-down role in settling these debates before they become larger problems.
2. Procedural Approaches Internal to the Executive Branch
A second set of procedural approaches could be implemented within the executive branch but external to HHS. Many initiatives in health innovation policy are run through the White House rather than through HHS, such as the above-described examples of the BRAIN Initiative, Personalized Medicine Initiative, and Cancer Moonshot. The President obviously has an interest in the success of these initiatives, and that interest may extend to associated collaborations between the agencies in question.
First, it is worth returning to the Rai and Benjamin proposal to establish an entity within the executive branch to promote innovation.251 Importantly, the Rai and Benjamin proposal is primarily directed at encouraging agency coordination, but it can easily be extended to encourage agency collaboration of the type I propose here. They consider the advantages of centralizing an innovation office and housing it within the executive branch, arguing quite rightly that the decentralization we observe at present is precisely the problem from an innovation perspective.252 Interestingly, in keeping with many of my arguments in Part III, Rai and Benjamin would create a regulator with both “an obligation and an incentive to operate transparently.”253
Rai and Benjamin would house their innovation regulator within the executive branch generally, rather than within a particular administrative agency.254 However, I have just suggested creating an officer for these purposes within HHS. In my view, my proposal may strike a middle-ground between complete centralization and complete decentralization. As compared to housing separate innovation officers within the NIH, FDA, and CMS or within the Executive Branch generally, housing an officer within HHS but giving them control over the collaborative activities of the subsidiary agencies is a compromise. This situation may allow the officer to develop deeper, more personal connections with the material than would housing them within the executive branch,255 while at the same time minimizing some of the difficulties of decentralization.
Second, the executive branch as a whole already has a system for requiring interagency consultation in particular cases: through the Office of Information and Regulatory Affairs (OIRA). OIRA is responsible for reviewing “significant regulatory action[s]” under development by administrative agencies,256 and in the course of that review, OIRA often facilitates interagency consultations. Professor Cass Sunstein, who was the OIRA Administrator under President Obama from 2009 to 2012, subsequently wrote that
[t]he governing idea is that relevant agencies have information and expertise, and the rulemaking agency should benefit from their perspectives before finalizing or even proposing rules. A central goal of the OIRA process is to ensure that rulemaking agencies have access to the wide variety of perspectives that can be found throughout the executive branch.257
When OIRA reviews agency regulations that implicate healthcare innovation, it might formalize this process of interagency consultation, although that may fall short of active collaboration.
Unfortunately, OIRA review will only capture a small subset of actions that are important for healthcare innovation. Much of the NIH’s priority-setting and grant-making activities would not qualify as “significant regulatory action[s]” warranting OIRA review. The FDA certainly makes rules, but more commonly it issues guidance documents that are not always258 subjected to OIRA review.259 To the extent that it is CMS’s possession of information rather than its rulemaking ability that contributes to much of the above discussion, that would not obviously implicate OIRA review either.
3. Procedural Approaches External to the Executive Branch
A third and final set of procedural options exists outside of the executive branch. Specifically, an interested Congress may choose to play a greater role in the development of innovation policy. This Article has considered a number of examples of congressional involvement in this area including the passage of 42 U.S.C. § 283a, requiring the NIH Director each year to report on the NIH’s collaboration with other HHS agencies260 and the creation of the Interagency Pain Research Coordinating Committee (IPRCC) and specification of its duties by statute.261 The executive branch has real institutional advantages in terms of flexibility and expertise, but Congress has tremendous power to force collaboration itself, require the transparency of that collaboration, and fund the collaboration.
The first and most obvious way for Congress to involve itself in health innovation policy would simply be to codify a number of the approaches suggested above. Such statutes could remain process-oriented in the IPRCC mold, sticking to specifying goals, committee composition, regularity of meetings, and reporting obligations, while avoiding substantive issues that would be worked out by the collaborations themselves. Congressional involvement might be particularly useful for collaborations requiring additional resources, as Congress can appropriate funds for projects it wants to foster.
But a second and potentially more generative solution may be to develop a non-partisan, independent organization along the lines of MedPAC, the Medicare Payment Advisory Commission.262 MedPAC is composed of independent experts who advise Congress on potential reforms to Medicare, broadly speaking.263 These experts are drawn primarily from academia, health care providers, and the health care industry more generally.264 Such a commission could not only review the individual actions of the NIH, FDA, and CMS as they relate to innovation policy, but it could also review the ways in which these agencies relate to each other.
This idea is not merely speculative. An early draft of the 21st Century Cures Act would have created a national Medical Product Innovation Advisory Commission based on MedPAC. The proposed Commission would “analyze medical product innovation in the United States and recommend policies to accelerate the discovery, development, and delivery of new medical products.”265 As such, the Commission would not only be specifically tasked with reviewing policies of the NIH (discovery), FDA (development), and CMS (delivery), but would also be tasked with “review[ing] the interaction of Federal agencies with respect to the discovery, development, and delivery of new medical products and how such interactions influence medical product innovation.”266 Unfortunately, this provision was removed from all subsequent drafts.
Conclusion
In this Article I have argued that scholars and policymakers ought to look beyond the capacities of administrative agencies like the NIH, FDA, and CMS individually to contribute to health innovation policy, and instead ought to focus on the potential for collaboration across agencies. Critically, there are important areas of collaboration which are almost entirely unexplored, and we ought to consider procedural options for encouraging or requiring such collaboration. The ultimate point of this Article, though, is broader. It presents a view of these agencies, which have distinct missions regarding overlapping subject matter, that is coherent and provides a way forward for innovation policy law and scholarship.