The Diversity of Institutions Conducting Biomedical Research

Abstract

Biomedical research in the United States has contributed enormously to science and human health and is conducted in several thousand institutions that vary widely in their histories, missions, operations, size, and cultures. Though these institutional differences have important consequences for the research they conduct, the organizational taxonomy of US biomedical research has received scant systematic attention. Consequently, many observers and even participants are surprisingly unaware of important distinguishing attributes of these diverse institutions. This essay provides a high-level taxonomy of the institutional ecosystem of US biomedical research; illustrates key features of the ecosystem through portraits of eight institutions of varying age, size, culture, and missions, each representing a much larger class exhibiting additional diversity; and suggests topics for future research into the research output of institutional types that will be required to develop novel approaches to improving the function of the ecosystem.

The United States leads the world in the size, scope, and impact of its biomedical research enterprise (MacLeod and Urquoila 2021). This results from many factors, including a skilled scientific workforce, substantial funding, and numerous institutions that have chosen to prioritize research within their missions. The scientific workforce and research funding have received substantial attention, including numerous inquiries into the economics of research (Azoulay and Li 2020; Partha and David 1994; Pickett et al. 2015; Stephen 2012). But—perhaps surprisingly—far less attention has been paid to the nature and characteristics of the institutions at which this research is conducted, despite the fact that their widely varying history, character, size, and missions are likely to be consequential for the research they conduct. Efforts to enhance the bioscience research ecosystem require that we understand and distinguish among these institutional homes for biomedical research.

Here, I present a novel institutional taxonomy followed by high-level portraits of eight institutions that hold biomedical research as integral to their missions. I examine their research from scientific, cultural, organizational, and financial perspectives. No two institutions are precisely alike, and neither this nor any other taxonomy can hope to perfectly capture the full institutional diversity. But most biomedical research takes place at institutions with substantial resemblance to those presented here. Thus, lessons learned from the proposed taxonomy and these examples will offer a useful roadmap for analysis of many similar institutions across the country. A major goal of this essay is to stimulate further refinements and clarifications of the institutional taxonomy.

The focus of this article is on bioscience research in nonprofit organizations, most but not all of which are linked to universities and schools. Most scientific advances arise from such institutions and their faculty; whatever their ultimate career destinations, essentially all researchers pass through them for education, training, and experience. Though they share many attributes, these institutions also differ in several characteristics. Of course, for-profit biopharmaceutical, device, and diagnostic companies play distinct and critical roles in biomedical discovery and most uniquely in therapeutic development, and there are many important interactions between the not-for-profit and for-profit domains. But nonprofit institutions are the specific focus of this essay.

Colleges and universities are longstanding homes for bioscience research. Beginning with a core mission of education, a subset of "research intensive educational institutions" began in the 20th century to adopt an additional goal, that of creating new knowledge through research. In many cases, these efforts grew to very large size. Most bioscience research at colleges and universities is conducted by faculty who also teach science to undergraduate and graduate students, some of whom aspire to scientific careers. Not surprisingly, the largest share of university bioscience research is conducted in graduate professional schools of medicine and public health, most of which are university owned or affiliated.

The landscape of institutional medicine is itself complicated. Medical schools must offer clinical apprenticeships, and this requires them to own or affiliate with hospitals. These hospitals may be owned by their parent medical schools and universities (such as those at Johns Hopkins, Penn, Duke, Stanford, UCSF, and others), or they may be independent corporate entities that operate under various "affiliation agreements" with the schools. This is the case at university hospitals affiliated with Harvard, Yale, and New York Presbyterian, the latter affiliated with two different medical schools, those of Columbia and Cornell. How research is situated and managed within these medical school/hospital complexes is also highly variable and is often unclear to external observers. Whereas research at Harvard-affiliated hospitals is "owned" by these hospitals and conducted in hospital-owned facilities by faculty with HMS appointments who are employed by the hospitals, most research by faculty at Yale, Columbia, and Cornell medical schools is "owned" by and conducted in school facilities, also by faculty with school appointments. Many school research faculty hold additional titles in the hospitals where they perform various non-research functions. Whether academic hospitals are university owned or independently governed, and whether research is owned/managed by the schools or the hospitals, these "academic medical centers," as they have come to be known, are the largest institutional engines of biomedical research, funded most importantly by the National Institutes of Health (NIH).

But academic biomedical research isn't limited to academic medical centers and their constituent hospitals and schools of medicine and public health. Such research also occurs at universities that have no medical schools (such as MIT, Princeton, UC Berkeley, Caltech) and at a wide variety of independent research institutes, some of which are university affiliated. The Association of Independent Research Institutes (AIRI) represents 90 such independent institutes, including Rockefeller University, HHMI, Cold Spring Harbor Laboratory, Scripps Research, the Salk Institute, and many others.

To structure the ensuing discussion, I have divided existing institutions into five major categories. Though these overlap in various respects, the resulting organizational taxonomy may be useful for considering the many hundreds of biomedical research organizations (Figure 1).

Academic Medical Centers

The first and largest category, as assessed by funding and number of faculty, are the academic medical centers, which for this purpose may be divided into three types. The first are like those at Hopkins, Stanford, Duke, Penn, and UCSF, where universities or medical schools own the hospitals, under integrated university governance. The second are academic medical centers in which schools and hospitals are affiliated but independent, and research by hospital-based faculty is predominantly conducted at or through the schools. The third are research-intensive hospitals that are independent corporate entities "affiliated" with schools/universities—such as Massachusetts General Hospital (MGH) and several other

Figure 1.

Proposed taxonomy of US not-for-profit biomedical research institutes

Harvard-affiliated hospitals—where research by hospital-based faculty is owned by and conducted at the hospitals, rather than the schools. In the latter cases, the school remains connected to hospital research via its role in approving faculty promotions and appointments, establishing policies on topics such as research integrity and conflict of interest, and conducting departmental reviews, among other activities.

Schools of Medicine

The second category includes a subset of schools of medicine (for example, Harvard Medical School) that neither own their affiliated hospitals nor the research conducted by faculty employed by these "affiliated hospitals." Consequently, research at these medical schools resembles, and is therefore included here with' research at universities such as MIT, Princeton, Berkeley, Caltech, and others that lack formal associations with medical schools or hospitals.

Disease-Focused Research Institutes

The third category includes specific disease-focused research institutes (variably engaged with patient care as well as research), the largest number being cancer research institutes, such as DFCI, Memorial Sloan Kettering, Fred Hutchinson, and MD Anderson. Most are independently governed but affiliated with medical schools.

Other Research Institutes

The fourth category includes research institutes without a specific disease focus, such as Rockefeller, Broad, Janelia, and Arc, as well as Scripps, Salk, and many others. Most are independent and some have university affiliations.

NIH Intramural Research Program

The fifth category is the NIH intramural research program, a very large government-based bioscience research institution with a broad remit and no university affiliation.

_______

Over the past century—in response to scientific opportunities, medical needs, and financial incentives—the biomedical research ecosystem has grown, both within existing and newly created institutions. This institutional diversity has been a major driver of the success of the system, but has created challenges as well as benefits. To plan and fund this critical ecosystem going forward requires that we understand these diverse institutional attributes.

To bring specificity to this analysis, I have selected eight institutions that span the five categories above. Choices were influenced by their ability to illustrate the proposed taxonomy and by the depth of my institutional knowledge from my time as an employee, leader, or adviser. To address the unavoidable trade-off between the value of deep institutional knowledge and the generalizability of the examples, a larger number of case studies are institutions where I had no personal experience. To further enhance generalizability of the examples, I have sought wherever possible to compare chosen cases with others in their class. Many other institutional exemplars might have been chosen to reflect additional geographic, organizational, or reputational axes of diversity, though no such choices would be perfectly representative. In the end, the eight selected institutions, each considered where possible in the context of others in their class and the larger national research ecosystem, are presented below.

The Harvard Medical School (HMS) Quadrangle

I begin this section with the Harvard Medical School (HMS), an institution that I led for nine years, during which I became intimately familiar with its highly complex organizational structure, which has similarities and differences to other medical schools. The HMS Quadrangle is a medical school–based, basic science–focused research program.

HMS is first and foremost a medical school, educating physicians since 1782 as a part of Harvard University, its dean reporting to the university president. The school implements the medical curriculum, run by educational leaders recruited to oversee it. The initial "preclinical" medical education takes place largely in medical school facilities, taught by a mix of faculty. Some are directly employed by the HMS educational program and the schools' basic science departments, though the role of the latter in medical education has diminished over time in essentially all schools. Many essential subjects, such as physiology and pathophysiology of disease, require faculty experts largely employed by HMS-affiliated hospitals, as I was when I taught diabetes and metabolism to HMS students for many years. Hospital-employed faculty involved in preclinical medical education are a tiny minority of the 12,000 (!) full-time HMS faculty employed by HMS-affiliated hospitals, many more of whom participate in hospital-based clinical education.

Despite the essentiality of medical education to a medical school, the lions' share of the $760 million-plus 2021 HMS annual operating budget supports research, not medical education. Research is directed by 170 or so faculty recruited and employed by Harvard, housed in HMS facilities, their research managed by HMS administration. Far more faculty with HMS appointments (several thousand) conduct research at affiliated institutions and hospitals (such as MGH, discussed below). In the HMS system, those faculty are employed by the affiliates, and their research is administered by these institutions, appearing on their budgets as opposed to that of HMS. This distinguishes HMS from most other research-intensive medical schools (Stanford, UCSF, Duke, UPenn, and many others), where research conducted by hospital-based faculty is largely administered by the school and/or university.

So, what characterizes the biomedical research carried out at HMS as opposed to its affiliated hospitals? Research by HMS-based faculty is performed in 11 "Quad departments" (the HMS campus is referred to as the Quadrangle), each with a department chair appointed by the dean. Eight are discipline-based "basic science" departments, such as biochemistry and molecular pharmacology, and microbiology; one—biomedical informatics—is in a realm of its own; and two—health care policy and global health and social medicine—perform research on social science related to medicine. These Quad faculty are Harvard employees with offices and labs in facilities owned by the school, supported by a mix of school and external funding. A modest number of faculty (varying by department) are employees of the affiliates with research programs located there, but academic appointments in Quad departments where they actively participate in scholarly and educational activities. Apart from the very limited number of such "affiliated faculty" within Quad basic and social science departments, the vast majority of HMS faculty who conduct research do so at the affiliated hospitals and institutes; these faculty have no academic association with Quad basic science departments, even when they share a research discipline, such as biochemistry or microbiology. While the HMS research ecosystem might be stronger if a greater number of hospital-based HMS faculty had meaningful associations with the Quad departments, this idea has not appealed to Quad department leadership, who see advantages in limiting the size of their communities.

The HMS Quad model for faculty recruitments is that of Harvard University. New hires are carefully controlled by the school's administration and require international searches with candidates chosen after advertisements that garner hundreds of applicants. The reasons for tight control are simple: each recruitment has substantial scientific and financial impact. Most recruited faculty enter at the assistant professor level on an 11-year tenure clock. Achieving tenure ensures commitment to financial compensation (for as long as the school and university remain solvent). Most recruits do achieve tenure, but the minority who do not must leave. To facilitate success, these highly vetted recruits receive substantial recruitment packages, totaling up to several million dollars to support lab start-up, personnel, supplies, and equipment. But even these funds last for a limited time. A sustainable research program requires successful competition for external research funds, and of course, success is also premised on external validation and recognition. Faculty are incented to offset salary through external funding. This approach at the HMS Quad is very similar to recruitment to basic science departments in most other medical schools, whether they are university-affiliated or owned.

A demographic consequence of this strategy is that the great majority of Quad faculty are tenured full professors, who tend to remain essentially indefinitely with committed salary support. The minority at lower academic ranks work hard to achieve tenure. Since most have traversed the rigorous process, faculty quality tends to be quite high.

The Quad department chairs are historically powerful, granted substantial independence to lead their departments. If there is anything resembling a research steering committee, it is the combined chairs of these departments together with the dean. The preclinical departments were originally established to support disciplines required for preclinical medical education (anatomy, physiology and pharmacology), but as biomedical sciences evolved, connections of these departments to the medical curriculum became increasingly tenuous. Departmental names changed over time (departments of anatomy, physiology, pharmacology, and pathology no longer exist at HMS, as departments such as cell biology, genetics, neurobiology, and systems biology came into existence). Research profiles of the newly named and configured departments have become less distinct (similar research often occurs in departments of biochemistry, cell biology and genetics). Details vary, but changes such as these in the identity and focus of basic science departments have occurred at most medical schools.

How successful is this Quad research model? Its relatively small and highly regarded research faculty garnered $176 million from the NIH in 2021. The school assumes responsibility for faculty salaries (while encouraging coverage on grants that permit this) and covers many departmental expenses, offering generous research space allocations assigned in proportion to external funding. Like virtually all other schools, HMS also oversees graduate degree programs in biomedical research. This model requires substantial sums from internal school resources. This fact became acutely evident in 2008, my second year as dean of HMS, when the US recession knocked 30% from the medical school's then $4 billion share of the university's $40 billion endowment, from which most unrecovered costs of research were paid. Many actions were required to reduce what rapidly became a substantial annual operating loss for the school.

With university agreement, I sought a major donor for a gift sufficient to justify naming the school or one of its illustrious programs, an increasingly common occurrence at medical schools (examples include Perelman at UPenn, Weill at Cornell, and Icahn at Mount Sinai). That effort came to fruition in 2018, after I had stepped down, with a $200 million gift from businessman and philanthropist Sir Leonard Blavatnik, for which the Quad-based research program was named the "Blavatnik Institute for Biomedical Research at HMS."

Research at the HMS Quad functions largely independent of that conducted at the school's affiliated hospitals, so its character resembles that at universities lacking medical schools or affiliated hospitals, like Princeton, Caltech, Berkeley, and MIT. MIT is linked to HMS through a subset of HMS medical student education (20%) called Health Science and Technology (HST), jointly run by HMS and MIT faculty. MIT also shares participation in the HMS MD/PhD program and with medically linked programs such as the Broad Institute, discussed in detail below.

HMS has for many years ranked atop the US News ranking for research intensive medical schools. This ranking reflects many factors, including selectivity of admissions, faculty/student ratios, faculty publications, student success, and various reputational indices. Research conducted by HMS hospital-based faculty is aggregated with Quad research for this ranking, which has annoyed its competitors but does reflect the overall reputation of the school, the great majority of whose faculty are employed by HMS-affiliated hospitals. The HMS model is quite uncommon among its reputational competitors and medical schools generally, where most faculty research is owned/managed by the schools, not the school-affiliated hospitals.

The nature of research at the HMS Quad—and at MIT, Princeton, UC Berkeley, Caltech, and other institutions like them—overlaps with but is clearly distinct in focus and scope from research conducted at academic hospitals, of which there are many hundreds. I have chosen Massachusetts General Hospital as the primary example because it is the largest such hospital-based research program and because I had the opportunity to observe and interact with it for many years.

Massachusetts General Hospital (MGH)

Massachusetts General Hospital (MGH) is the oldest and largest of 17 institutions affiliated with (but not owned by) HMS. As such, it represents a medical school–affiliated but independent hospital research program. Founded in 1811, MGH was the third general hospital in the US, admitting its first patient in 1821. The hospital was initially intended to care for the poor in an era when those with sufficient means preferred medical care at home, even as most of that care, as we now know, was ineffective or harmful. Two hundred years later, MGH has 1,000 inpatient beds, admits about 50,000 patients, handles 1.5 million outpatient visits, and performs over 34,000 surgical operations each year. With more than 25,000 employees, MGH is big business, the largest nongovernment employer in Boston, with a 2018 annual operating budget of $4.5 billion. In recent years, MGH has been ranked highly in Best Hospital rankings by US News & World Report.

In addition to its core mission as a hospital, MGH is also the largest hospital-based research program in the world. Although the scale of research at the 17 HMS affiliates varies widely, several are extremely large (Brigham and Women's, Boston Children's Hospital, Beth Israel Deaconess, and Dana Farber Cancer center) though smaller than MGH. With a research budget of $1.2 billion in 2021, MGH research takes place in more than 1.3 million square feet of interior space, across many buildings and sites. Growth required a large secondary research campus at the former Charlestown Navy Yard, about two miles from the hospital campus. Research is conducted by faculty in more than 30 departments, centers, and institutes, including six "thematic research centers," such as those for regenerative medicine and for genomic medicine. Trans-departmental research centers have become more common in many hospitals. As elaborated below, though MGH is the oldest and largest HMS affiliate, and all its faculty have HMS appointments, its research program is owned and managed by MGH, operating in accordance with a variety of HMS policies and guidelines.

Who conducts the research? Overall, MGH has about 5,000 full time faculty, all holding HMS appointments, the largest single contribution to the over 12,000 full-time HMS faculty. About 2,000 of the approximately 5,000 MGH faculty are MDs, PhDs, and MD-PhDs whose primary job is research, leading research teams as "principal investigators" (PIs). In addition to these 2,000 research faculty PIs, about 1,000 of the hospital's 3,000 practicing physician faculty participate in clinical trials, 2,000 of which are ongoing at any time.

The success of biomedical research and care at the affiliates benefits greatly from the HMS appointments of their faculty, and this should not be underestimated. HMS policies and the Harvard imprimatur ensure a consistent high academic standard across the affiliates. The association with Harvard increases the attraction of faculty to holding positions at the affiliated institutions, as do university affiliations of other leading academic hospitals, whether they are university owned or have independent governance.

Research at MGH ranges from highly basic to clinical in every conceivable specialty and subspecialty. MGH PIs publish an average of 500 papers per month in peer-reviewed journals. Beyond these 2,000 faculty working as research PIs are 4,000 non-faculty scientists who work with and support them, 1,500 postdoctoral fellows, many graduate students (some MGH faculty are granted access to bioscience PhD students at HMS/Harvard University), and 1,200 administrative and support staff. This total research workforce of 9,500 comprises more than a third of all hospital employees.

What financial sources enable this billion-dollar research program? The biggest source—50% of the total—is from the NIH. More than 1,000 NIH grants to MGH—each submitted as an application by a PI and funded after competitive NIH peer review—provided more than $600 million in 2021. This is the tenth highest amount of NIH funding to any US institution, more than any other independent hospital, and exceeding that of most universities; Stanford University, whose total NIH funding includes that of its hospital, has about the same NIH funding as MGH. (When combined, annual NIH funding to HMS and all its affiliated institutions exceeds $2 billion.) NIH support is supplemented by funding from assorted nonprofits and foundations, which contribute $175 million or 15% of the total, with industry-sponsored research providing 7%, or another $83 million.

At MGH and elsewhere, external research funding comes in the form of both "direct costs" for supplies, salaries, and equipment controlled by the PIs and "indirect costs" (IDC) paid to the institution to cover facilities, administration, and related expenses not charged to individual faculty. MGH counts the shortfall—those institutional costs not covered by external payments—as $60–80 million per year, which must come from "internal sources." The precise provenance of these funds is difficult to pin down, and how they are accounted for is a common topic for debate and discussion at MGH and elsewhere, but there are general outlines. Some comes from financial margin and accumulated reserves of clinical departments, which vary greatly by specialty. Philanthropy to support research is a major source of overall philanthropy at research-intensive hospitals, and MGH has been very successful in this regard. For example, software entrepreneur Phillip Ragon and his wife Susan Ragon have given $400 million to MGH over the past 12 years to support the Ragon Institute of MGH, MIT, and Harvard, based in Cambridge and managed by MGH. Its focus is on the immune response and disease, including vaccine research and development.

Another source of institutional research funds derives from revenues from patents, licensing, venture capital backed startups, and other commercial activities related to research. MGH has had substantial success in this area, which rarely develops as a reliable mechanism to support hospital research. An MGH researcher, Brian Seed, was the inventor on a key patent held by MGH that enabled development of the blockbuster drug Enbrel, a mainstay of treatment for rheumatoid arthritis. This patent has brought MGH over $600 million in royalties and related revenues. Prevailing revenue-sharing arrangements disbursed substantial funds to Dr. Seed personally, to his lab and department, and to the hospital.

How is research organized and managed within an institution simultaneously responsible for safely and effectively delivering diverse medical services? MGH is a nonprofit organization overseen by an unpaid board of directors.¹ Reporting to current CEO David Brown, an emergency medicine physician, are executives responsible for corporate finance, operations, space, human resources, philanthropy, and other areas. He and a physician executive for the "physician's organization" jointly oversee chairs of 30 clinical departments, such as medicine, surgery, obstetrics and gynecology, neurology, radiology, pathology, and many others. All MGH faculty require both HMS and hospital appointments, with the hospital fully responsible for their employment and compensation packages.

Research decisions were once solely the province of departments and divisions, some of which became research powerhouses. That distributed approach works less well in a program of this magnitude and scope, though this approach remains in many other hospitals. Over the past 15 to 20 years, an overarching research management infrastructure developed to interface with department-based programs, which still play a dominant role in shaping the faculty. An MGH Research Institute is overseen by the CEO and chairs of medicine and surgery, to which a Research Institute Steering Committee of faculty and administrators reports. The institute has a scientific director and director of clinical research, who focus on hospital-wide research issues. This structure also enables better "branding" of the MGH research program, which all such institutions seek, in part to encourage philanthropic contributions. An executive committee on research (ECOR) has elected members representing the broader faculty, and the HMS dean sits on its advisory board. ECOR is responsible for overall policies and distributes funds provided by the hospital to support investigators via a variety of initiatives. A senior vice president for research, Harry Orf, oversees research operations. But despite this elaborate administrative superstructure, research at MGH is hardly a centralized endeavor. As described to me by Orf, the approach is largely one of "let a thousand flowers bloom," as explained below.

How are MGH researchers recruited, developed, incentivized and rewarded, or—in management speak—what is the "talent model"? As in most hospital-based research programs (and as distinct from most universities and medical school basic science), most MGH faculty come up through the ranks, moving from their positions as MD and PhD trainees (largely within clinical departments) to junior faculty positions based on their desire to continue research, their demonstrated early success, an existing lab wanting to retain them, and eventually their ability to garner external research funding, often initially through trainee transitional grants. The high quality of MGH trainees enables this approach to succeed, and though the transition from trainee to faculty member is often difficult, there are many successes. A minority of faculty are recruited from outside MGH through open academic searches led by MGH in coordination with HMS. Whether hospitals are governed independently like MGH or university owned, the pattern of most faculty in hospital-based departments coming up through the ranks is typical.

The vast majority of research faculty, including those in senior positions, operate as small business entrepreneurs, the sustainability and growth of their programs largely dependent on their ability to obtain funding, what some refer to (perhaps unappealingly) as the "eat what you kill" approach. Few MGH faculty have anything approaching financial tenure. If external funding disappears, the great majority—even those in senior positions—see their research programs diminished or ended, potentially mitigated by periods of bridge funding provided by departments, ECOR, and supportive donors. Those physician-scientists who are willing and able may take on clinical activities to fund their salaries and assume administrative roles.

When I asked Harry Orf what keeps him up at night about the MGH research program, his answer was instantaneous: "It's all soft money." This enormous program with its dedicated space, research infrastructure, administrative support systems, and support for salaries and supplies depends on essentially continuous receipt of external funds, requiring faculty and leaders to devote enormous time and effort to raising funds. This occurs amidst uncertain NIH budgets and changing research priorities, an uneven economy, and unpredictable philanthropy and research business ventures. Orf is confident but not relaxed, nor should he be.

To compare the enormous MGH research program to that of the HMS Blavatnik Institute for Biomedical Research, both share high quality and recognition, with faculties of similar training and skill that pursue many overlapping scientific questions. But they differ in several key respects. The Quad program has far fewer faculty, and their numbers and disciplinary distribution are tightly controlled. HMS places bigger bets on a smaller number of people. Quad faculty research is much more "basic" in its orientation, and far less commonly directly addresses the medical disorders that typically occupy the larger, more scientifically diverse and clinically oriented MGH faculty. Unlike at MGH, very few Quad faculty split their efforts between clinical activities and research. And finally, facing an up-or-out situation, Quad faculty are given substantial initial resources ("recruitment packages") to enable research sufficient to meet the high bar for professorial tenure as judged by their departments, HMS, and University leadership. HMS commits to their salary (but far less to research operational funds) if grants are not obtained.

In contrast, the vast majority of MGH and other affiliated hospital research faculty (and faculty in clinical departments at other medical schools) do not face an up-or-out tenure clock, are not granted financial tenure even at the professorial level and can remain at lower academic ranks indefinitely. Many receive limited (or no) start-up packages and must compete from the beginning for sufficient resources (salary and operational) to succeed. This model wherein most junior faculty begin in the labs of more senior faculty diminishes the requirement for large recruitment packages. Salary commitments from the institution are highly variable and individualized, and many have no long-term commitments. Perhaps this comparison is best seen—to build on the Orf analogy—as a limited number of highly cultivated orchids (at HMS and other basic science departments) versus a thousand blooming wildflowers (at MGH and most other hospitals with research in clinical departments).

At the great majority of academic hospitals, research exists on a far smaller scale than MGH, though many do have research efforts of substantial size. The constant challenges to these hospitals relate to largely soft-money (as opposed to institutional-derived) research funding and the need to balance the costs of research with necessary investments in their primary clinical missions. Research in all hospital-based clinical departments is extremely vulnerable should hospital clinical operating margins diminish.

The large size of research programs at MGH and the other HMS-affiliated but independent hospitals (including Brigham and Women's, Beth Israel Deaconess, Boston Children's, and others) is notable. Outside the HMS ecosystem, most of the largest hospital-based research programs are in hospitals owned by their universities, such as those at Johns Hopkins, Stanford, Penn, Duke, and UCSF. At those institutions, it is difficult to parse the research associated with hospitals versus their schools and universities, and public funding figures often aggregate these.

It is worth considering whether research in hospitals like MGH differs from that in hospitals owned by and therefore more deeply embedded in their parent universities. Research at UCSF has basic science research and research in hospital-based clinical departments all residing within the same governing organization. Research at UCSF has been extensively examined in a book-length treatment that describes many tensions between "basic science" research departments and clinical departments in an organizational model with centralized governance (Bourne 2016). Further research on this topic should be conducted.

The Dana Farber Cancer Institute (DFCI)

Disease-focused research institutes have characteristics distinct from those of medical schools and hospitals, and therefore merit consideration as a separate category in this taxonomy. Cancer research institutes are the most prominent among these. The Dana Farber Cancer Institute (DFCI) is one of several cancer research institutes in the US, several of which are among the biggest homes for biomedical research and NIH funding. These include MD Anderson in Houston, the Fred Hutchison Cancer Center in Seattle, and Memorial Sloan Kettering in New York City. Each of these has independent governance and each raises enormous amounts from philanthropy.

DFCI is a major independent affiliate of HMS, the anchor of the NCI-funded Dana Farber Harvard Cancer Center. With an annual budget of $2 billion and an endowment of $2.3 billion, DFCI faculty conduct more than 250,000 outpatient visits annually; inpatients are seen by DFCI clinical faculty at Brigham and Women's Hospital or Boston Children's Hospital. DFCI employs more than 550 faculty with HMS appointments, of which more than 250 serve as PIs on NIH grants, garnering NIH funding of $159 million in 2021. Both fundamental and clinical research related to cancer are central to DFCI's mission, and it oversees more than 700 active clinical trials. Building upon the tradition of philanthropy established by founder Sidney Farber, DFCI raised $403 million dollars in 2021, ranking it one of the largest cancer charities in the US.

The Fred Hutchison Cancer Research Center in Seattle, which merged with Seattle Cancer Care Alliance, Seattle Children's, and University of Washington Medicine in early 2022, recently announced a $710 million gift from the Bezos Family Foundation, one of the largest single gifts to a cancer research organization. Each of these organizations have affiliations of some kind with a local university and medical school. DFCI is the smallest of the institutions mentioned, and clinical care and research hold an approximately equal share of its budget.

As with MGH, all DFCI faculty hold HMS appointments. These originate within the academic programs of DFCI but are routed through HMS "appointing" departments elsewhere in the HMS system (for example, medicine through Brigham and Women's, pediatrics through Boston Children's). A few of DFCI's more basic scientists have academic appointments through HMS Quad basic science departments but are housed at and paid by DFCI.

One aspect of cancer clinical care and research at HMS and its affiliated institutions, as well as elsewhere in the country, is the often-intense competition among institutions for patients, reputation, federal funding, and philanthropy. While DFCI is the single largest and highest ranked cancer center in the Harvard system, each of the general hospitals has a substantial and excellent program in cancer clinical care and research, and with a couple of exceptions, these are independent of DFCI. Faculty at each institution receive their own cancer-related research grants. One tension relates to the fact that Harvard institutions have been permitted by NCI to hold only a single NCI Comprehensive Cancer Center Grant. These NCI grants support several important activities, but their financial magnitude is less than the prestige and influence they confer. This center grant was originally submitted and long held by DFCI, but when MGH expressed an interest in applying for their own such grant in the early 2000s, NCI informed the HMS dean that only one would be permitted for the Harvard system. After extensive negotiation, it was agreed that the grants administrative home would remain at DFCI, but its name would be changed to the Dana Farber Harvard Cancer Center (DFHCC), with the HMS dean chairing the governance committee. As the MGH Cancer Center grew in size and influence, I was repeatedly implored by MGH leadership to change the name to the Harvard Cancer Center, rotating administrative leadership among the major Harvard institutions. The politics of this struggle was intense. During my tenure as dean, we tried to enhance the already substantial cross-institutional collaboration, but the name and governance of the DFHCC were not altered.

Similar issues have arisen with other disease–specific research institutes. The Joslin Diabetes Center (JDC) is a venerable institute for diabetes research, education, and clinical care located adjacent to HMS in the Longwood Avenue area. Founded in 1898 by Dr. Elliott Joslin, the JDC sees diabetes outpatients at its own facility and shares management of inpatient care with its clinical affiliate Beth Israel Deaconess Medical Center (BIDMC). Most recently, JDC has become a part of Beth Israel Lahey Health, the parent of BIDMC, so its governance is no longer fully independent. Thirty HMS faculty employed by Joslin and more than 300 scientists working with them conduct research in all areas related to diabetes, from basic through clinical. While Joslin is the best-known center for diabetes research in Boston, each Harvard-affiliated hospital also has a substantial diabetes research portfolio, and they have preferred to retain these rather than have Joslin serve as Harvard's primary diabetes research center.

Aspects of these two Harvard examples are replicated in many other academic medical centers, and they illustrate the tensions between collaboration among researchers, wherever employed, and competition between and among institutions for patients, reputation, and funding from government, philanthropists, and other sources. Most institutional leaders understand this tension and try to mitigate it, but competitive institutional instincts and bureaucratic roadblocks not infrequently impede collaborative efforts of their research faculty. Nonetheless, most scientists believe that collaboration advances their research, and many collaborations—across Harvard, nationally and internationally—pay little heed to institutional boundaries.

Rockefeller University

Rockefeller University (RU) is the oldest independent biomedical research institute in the US. Founded in 1901 as the Rockefeller Institute for Medical Research by industrialist and philanthropist John D. Rockefeller, it has been located since 1906 on a large campus on the Upper East side of Manhattan. RU is a private, independent, graduate-only university (PhD degree candidates first admitted in 1955), with an endowment recently valued at $2.3 billion dollars. The Institute's founding charter stated its mission with great clarity: to develop a scientific understanding of "the nature and causes of disease and the methods of its treatment, and to make knowledge relating to these various subjects available for the protection of the health of the public and the improved treatment of disease and injury." This mission was far ahead of its time, and by any standard, Rockefeller has been extremely successful, its faculty making a steady stream of important discoveries. These include many advances related to infectious diseases in its early decades, and in 1944, the discovery by Avery, MacLeod, and McCarty that DNA is the substance that transmits hereditary information. RU faculty have received a disproportionate number of awards and recognitions: five current faculty members hold Nobel Prizes, eight have Lasker Awards, 31 are appointed as Howard Hughes Medical Institute (HHMI) investigators, and many have been elected to prominent honorary societies.

This success emerged from an organizational model that despite some changes over recent decades remains fundamentally different from most other biomedical research institutions (Hollingsworth 2002). Medical schools and academic hospitals generally organize research around departments with a disciplinary base, either basic or clinical. In contrast, the Rockefeller is organized around laboratories headed by individual scientists (approximately 70 heads of labs), each reporting to the president. By design, RU faculty do not hold appointments in nearby universities or medical schools, though the institution now collaborates with Weill Cornell and Sloan Kettering Cancer Center on PhD and MD/PhD programs. Rockefeller is led by a president, always a prominent scientist, selected by and under the direction of a fiduciary board of lay members and prominent scientists. With an overall annual research budget of $228 million, Rockefeller faculty received 145 NIH grants totaling $76 million in 2021, along with substantial financial support for 31 HHMI-funded faculty (an unusually high fraction of the faculty). This is supplemented by substantial institutional endowment funds. In early decades, all research was funded from the Rockefeller endowment.

The success of RU may be best explained by a "culture of excellence," coupled with a flat and nimble organizational structure that allows the assessment of talent and the ability to quickly deploy resources to recruit and retain that talent. When leadership recognized about two decades ago that its faculty demographics were aging, they decided to recruit more junior scientists via open searches, with a tenure-track path to ongoing support for salary and research. These open searches are broad and typically do not specify research disciplines or topics. Recommendations by faculty search committees are passed to the president, and if accepted, to a highly influential scientific committee of the board.

In addition to laboratory research, founding documents mandated that a research hospital be an essential component of the institute—once again, far ahead of its time. From its opening in 1910, Rockefeller University Hospital pioneered as a hospital geared towards research, with faculty seeking links between clinical practice and laboratory science. This is distinct from the more common history of hospitals adding research faculty and labs to their preexisting clinical mission (such as at MGH). Initially supported entirely by institutional funds, the Rockefeller Hospital program received funding from the NIH General Clinical Research Center Program in 1963, and since 2006, Rockefeller has held a Clinical and Translational Award (CTSA) from the NIH. The facility has 20 inpatient beds, 11 outpatient rooms, and support services enabling clinical research. Research is carried out in areas of clinical medicine and translational science of interest to the faculty.

Despite the absence of departments, and with research based in individual labs that are free to chart their own course, the institute has developed 10 or more interconnecting research areas and centers in recent years (including human immunology, cancer biology, and neuroscience) with which faculty may choose to associate. Driven by scientific affinities and philanthropists who fund and incentivize them, these associations promote cross-laboratory interactions and discipline-related community resources.

Given its success, could or should the RU model be further scaled? RU faculty and leadership seem to believe that faculty size is now roughly optimal; if the opportunity arose to substantially grow the faculty, several told me they would likely reject this opportunity.

Among many other nonprofit bioscience research institutes with broad research portfolios, Scripps Research and the Salk Institute deserve mention. Scripps Research began in 1924 in La Jolla, California, as the Scripps Metabolic Clinic, with a gift from philanthropist Ellen Scripps, and evolved through many stages to the creation of the Scripps Research Institute in 1993, recently renamed Scripps Research. More than 200 faculty are associated with five departments (chemistry, immunology and microbiology, integrative structural and computational biology, molecular medicine, and neuroscience), and several centers (including HIV/AIDS and metabolomics), and since 1989 it has had a highly regarded graduate program. Scripps is known for successfully merging basic biological science with chemistry, computational science, and translational science, emphasizing the emerging field of personalized medicine. Current faculty includes three Nobel Laureates and many recipients of prestigious prizes. Scripps's discoveries have led to many marketed medicines; translation of basic discoveries to therapeutic use is a major institutional goal. In 2021, Scripps faculty and programs held 233 grants totaling $207 million in NIH grant funding.

The Salk Institute for Biologic Studies, a neighbor of Scripps in La Jolla, was founded by polio vaccine discoverer Jonas Salk in 1960, on land donated by the City of San Diego, with initial funding from what is now the March of Dimes Foundation. Initial areas of focus were molecular biology and genetics, neurobiology, and plant molecular biology, with research units focused on diverse topics including cancer, neurobiology, immunology and microbial pathogenesis, gene expression, and others. Its 53 faculty have adjunct appointments at University of California San Diego, with which they share a bioscience graduate program. Like Rockefeller and Scripps, their faculty has included many Nobel Laureates and recipients of the most prestigious awards and recognitions. In 2021, Salk faculty held 99 grants totaling $61 million in NIH grant funding. Both Scripps and Salk have substantially lower endowments than Rockefeller.

RU, Scripps, and Salk are three examples of independent bioscience research institutes that, despite different organizational models, size, and geographical location, have shared success at the highest levels of accomplishment. Many other independent biomedical research institutes of variable size and quality exist and make valuable contributions to the research ecosystem.

The Broad Institute

Founded in 2004, the Broad Institute arose in response to the rapidly evolving scientific landscape in human genetics and a perception that potential benefits of cross-institutional collaboration were yet unfulfilled. Its creation required an outstanding and charismatic scientific leader (Eric Lander), wealthy donors excited by and committed to his vision (Edythe and Eli Broad), and existing institutions and faculty (from MIT, Harvard, and Harvard-affiliated hospitals) who agreed to engage in the new venture.

An MIT faculty member, Lander led its genome center that played a major role in the Human Genome Project, in the process acquiring management skills and approaches to scaling of research. Lander shared a vision with the Broads on how genome studies relevant to human health might be accelerated. He believed this would most efficiently develop by involving existing faculty from local institutions whose prior track record for inter-institutional collaboration was limited. Forceful leadership and philanthropy drove creation of the Broad, whose governance has evolved over time. Initial fiduciary control by MIT and Harvard (itself a quite unusual pairing) was followed by fiscal and fiduciary independence, under a board that now includes but is not controlled by leaders from MIT, Harvard, Harvard Medical School and its hospitals, along with other scientific and philanthropic luminaries.

Between 2004 and 2013, the Broad Foundation donated $700 million, $400 million of which contributed to its current $955 million endowment that enabled its quest for fiscal independence. Broad had operating revenues of $564 million in 2021, with 40% from federal sources including $171 million from NIH grants, $134 million from foundations and nonprofits, and $114 million from industrial sources. The Broad campus of three modern buildings is in proximity to MIT and is organized around three functional components: faculty, research programs, and platforms.

All Broad faculty hold appointments at MIT, Harvard, or a Harvard-affiliated teaching hospital, where they are expected to be "full citizens," though the implication of that citizenship is sometimes disputed. Broad-associated faculty parse into several categories, a complex organizational approach reflecting how the community was built over time. Faculty categories include institute members (core and non-core), associate members, and affiliate members, more than 400 overall. Institute members are the most engaged and set the scientific directions for the institute. Fifteen core institute members have their primary labs at the Broad (this category did not exist in its early years), and 51 non-core institute members have labs at home institutions while they also engage in or lead projects at the Broad. Associate members are (variably) active participants in the Broad community—attending some scientific meetings, leading or collaborating on one or more projects, and being eligible for internal seed funding. Some have laboratory space at the Broad. There are currently more than 300 associate members. Affiliate members have the least significant interactions.

Broad leadership initially built this community with faculty previously recruited and overseen by the affiliated institutions, rather than with new faculty directly recruited to the Broad. Some have viewed this as cherry-picking of faculty from these other institutions, while others consider it to have been a necessary approach to building otherwise difficult to establish cross-institutional programs. I see creation of the Broad having elements of both approaches.

In addition to individual labs of institute members, the Broad is known for many disease-related programs (such as cancer or diabetes) and programs that integrate cross-cutting scientific areas (such as drug discovery and medical and population genetics). As initially envisaged, many of these programs have become vibrant scientific communities. Several research platforms, including genomics, imaging, and proteomics, also play important roles in Broad culture and operations, supporting all research activities. The scientists working in and leading these platforms comprise distinct and important scientific communities. They are treated with a high level of professional respect by their colleagues, which enhances the quality of their work, an approach that might be more widely replicated.

The alliance of a forceful leader and extremely supportive donors operating in a rich scientific environment enabled the rapid ascent of a new organization that has added importantly to the local and national biomedical research ecosystem. It has done so both through research in individual labs, and via large-scale collaborative and interdisciplinary programs in areas like population genetics that likely require Broad-like scale and infrastructure to achieve their full potential. It's not clear that any other such Broad-like organizations exist and might be recreated elsewhere. One possible analogue is the Harvard-associated Wyss Institute for Biologically Inspired Engineering. Fueled by over $600 million in gifts from Hansjorg Wyss to Harvard University, the Institute has created productive cross-institutional collaborations both within and beyond the Harvard community, enabled by substantial technical and business development infrastructure and a unique approach to faculty and governance.

Janelia Farm of the Howard Hughes Medical Institute (HHMI)

The Howard Hughes Medical Institute (HHMI) is the second largest US nonprofit institution focused on biomedical research, with an endowment of $23 billion, exceeded only by the Bill and Melinda Gates Foundation. HHMI was founded in 1953 by American businessman Howard Hughes to support biomedical research. His death in 1976 and the subsequent sale of Hughes Aircraft Company generated substantial cash assets. The governance of HHMI then evolved towards greater professionalism, embodied in leadership by a group of outstanding scientists. The largest share of its annual budget supports approximately 250 scientists from 60 institutions across the country, each chosen after intense examination of their capability to pursue important scientific questions—their institutional philosophy is "people, not projects." Once selected, HHMI faculty receive substantial funds for salary and research at their home institutions for renewable terms of seven years or more.

In 2006, HHMI took on an additional goal, opening a new research campus in Ashburn, Virginia, known as Janelia Farm. Initially based loosely on Bell Labs, the outstanding basic research facility owned by AT&T from which nine Nobel Prizes arose, Janelia may be viewed as an experiment in creating a new scientific culture. The plan was to choose an area of biomedical research—initially neuroscience—and then to hire a group of outstanding scientists to work collaboratively in that area. To lead the faculty recruitment process, HHMI leadership appointed an initial group of highly respected senior HHMI scientists, one appointed as chief scientific officer. There was an additional emphasis on creating and deploying new enabling technologies—such as advanced molecular imaging, new sensing systems and others—to catalyze breakthrough discoveries that would then be taken up more broadly.

Janelia is modest in size, with about 40 faculty divided between group leaders and senior group leaders. Typical academic titles are not used, and Janelia has no university affiliation or associated PhD program. Faculty are encouraged to work actively in their own labs, overseeing smaller research teams than are typical for top academic scientists, who often tend to be judged in part by the size of their lab groups. Group leaders are also encouraged to collaborate with other labs. All research expenses are provided by HHMI, relieving faculty of the need to raise external funds, a major time sink and distraction in other academic settings. There is no long-term tenure track or commitment. After a review of accomplishments at seven years, scientists may remain at Janelia for an additional term or be provided resources to pursue careers elsewhere, some as HHMI investigators. A recent strategic review concluded that, after the first 15-year period working on neuroscience, other topics would be chosen for the second 15 years. These now include 4D cellular physiology and mechanistic cognitive neuroscience.

Janelia is, in some senses, an experiment in institutional design. Whether Janelia benefits from the lack of university affiliation or tenure, and its relatively remote location, are questions deserving attention. It would be of interest to determine whether and how the chosen design has influenced the quality and quantity of the research undertaken there.

The Arc Institute

Given the number and diversity of nonprofit institutions where biomedical research is conducted, it might surprise some observers that new varieties continue to be launched. But that is the case, and some of the motivations for doing so are interesting to consider.

In that light, consider the Arc Institute, launched in December 2021. Patrick Collison—a very wealthy young high-tech entrepreneur—developed an interest in biomedical research and questioned whether our substantial national investment produces less innovation than it might. He became convinced that dependence on NIH funding creates many adverse incentives, and this, together with other cultural factors, limited creativity, collaboration, and the ability to leverage technology to advance research.

Collison and a few others committed $650 million to launch the new, independent Arc Research Institute in Palo Alto, California, its focus being to conduct outstanding fundamental biomedical research. While being attentive to the problems of common, complex diseases, such as cancer and neurodegeneration, it would encourage its faculty to pursue any research questions they considered scientifically important. They would recruit about 20 scientists with a range of interests and accomplishments into newly renovated facilities, where each would run a lab fully funded by institute funds (like Janelia), for initial terms of eight years, potentially renewable. Faculty would be chosen for their potential to be innovators and encouraged to pursue their scientific curiosity, rather than committing via detailed proposals to specific hypotheses with identified deliverables (as NIH funding typically requires). With these core faculty as the first phase, they are adding technology cores in areas such as genetic technologies, in vivo model development and computation, and eventually a capacity to accelerate translation of selected basic insights into therapies.

Unlike the founders of Janelia (but like those of the Broad), Collison concluded that Arc's success required location close to an outstanding bioscience research community. Collison chose the Bay area (after considering the Boston/Cambridge hub). This reflected his belief that the most effective research community would include outstanding graduate students, whose youth and openness to new ideas might catalyze transformative discoveries. After sustained negotiation, Arc concluded agreements with Stanford, UCSF, and Berkeley, by which PhD students could choose an Arc lab for their thesis research, funded by Arc. In addition to core faculty located on the Arc campus, they anticipate a larger faculty community will emerge at academic partner institutions to engage with the Arc, like associate faculty at the Broad institute.

These ambitious goals will require wise choices of core faculty and technology platform leaders. The rapid success of the Broad owed much to the scientific judgment and leadership of Eric Lander, who had a proven ability to organize and lead cutting-edge genomic science and to motivate and support others. Stimulated by his vision and substantial philanthropic funding, Lander convinced several talented young scientists to join him in defining, initiating, and building the organization. This is not the initial approach taken by Arc, which named a highly accomplished but early-career scientist as its initial director, with authority to select new faculty, independent of any academic structure. When I asked Collison about this unconventional approach, he answered that, in his opinion, most organizations choosing academic scientists rely too much on process and committees, with excessive control by senior scientists. He believes outstanding young individuals with superb taste and judgment may produce equivalent or better recruiting outcomes. We agreed that the success of this unconventional approach will be determined by the institute's future scientific accomplishments, and even then, it will be difficult to judge success of contrasting approaches absent an experiment or formal analysis of some kind, which I hope is conducted. Whether this approach proves superior or not, the investment adds to the scientific enterprise, and institutional experiments of this nature should be welcomed, especially if their success is rigorously assessed.

The NIH Intramural Research Program

Most people know of the NIH as the federal agency within the Department of Health and Human Services (DHHS) that is the largest funder of biomedical research in the US. But few understand that 10% of the NIH's $43 billion budget funds an extensive "internal' research activity, its "intramural research program" (IRP), which is carried out on an expansive Bethesda, Maryland, campus and several other smaller venues. Any consideration of US biomedical research institutions must include this program.

The roots of the NIH go back to 1887, with the creation of a one-room laboratory within the Marine Hospital Service (MHS), the agency that preceded creation of the US Public Health Service (PHS). The MHS in turn had been established in 1798 to provide medical care to merchant seamen. The Treasury Department collected 20 cents per month from the wages of each seaman to cover treatment costs at a series of hospitals. In the 1880s, the MHS was charged by Congress with assessing arriving ships for passengers with signs of infectious diseases, to avoid dreaded epidemics of cholera and yellow fever. In 1891 the Hygienic Laboratory, as it came to be called, moved to Washington, DC, and in 1902 the Laboratory became a center for research within the federal government, reorganized and renamed as the Public Health and Marine Hospital Service (PH-MHS), conducting research in pathology, bacteriology, chemistry, pharmacology, and zoology. Authorized to hire researchers with PhDs in addition to physicians, its name was shortened to the Public Health Service in 1912, and its mission expanded to include research into noncontagious diseases. The National Institutes of Health, as it is known today, was established by the Ransdell Act of 1933. The National Cancer Institute (NCI) was the first categorical institute, formed in 1937—supported unanimously by the Senate—foreshadowing the creation of additional institutes.

During World War II, the NIH conducted substantial research in support of the war effort. As the war was drawing to a close in 1944, the NCI was formally made part of NIH by the Public Health Service Act, setting the stage for enormous postwar growth of the NIH, stimulated importantly by the highly influential report written by Vannevar Bush, Science, the Endless Frontier (1945). By far the greatest growth of the NIH budget was that for funding external research; that budget grew from $4 million in 1947, to $100 million in 1957, to $1billion in 1974, and to $43 billion today. The NIH Clinical Center, the first of its kind research hospital (apart from Rockefeller), opened in 1953 with more than 200 beds in close proximity to research laboratories. I spent four years in one of those labs from 1974–1978.

Today, the NIH IRP has a budget of $4 billion and conducts basic, translational, and clinical research across 50 buildings on several campuses, with the largest in Bethesda. There are about 1,200 principal investigators and 4,000 postdoctoral fellows in more than 20 institutes and centers, each of which runs its own programs within budgets established by Congress. The NIH web page says that based on its budget, the IRP is "the largest biomedical research institution on earth." The golden age of the IRP likely coincided with the Vietnam war, when the NIH attracted a remarkable cadre of physician-trainees—the "Yellow Berets"—who could avoid military service by gaining highly competitive positions in NIH labs. Many leaders of biomedicine emerged from that cohort (Azoulay, Greenblatt, and Heggeness 2021).

IRP scientists pursue both fundamental scientific questions and research into a broad array of diseases, including clinical studies in the Clinical Center research hospital. Research is organizationally divided into 27 institutes and centers, the three largest being the NCI, the Institute of Allergy and Infectious Diseases (NIAID), and the National Heart Lung and Blood Institute (NHLBI). Each institute has a director of intramural research who reports to the overall institute director, working as well with an NIH-wide director of intramural research. There are laboratories headed by PIs with a basic science focus, and branches and sections that have disease or clinically oriented research directions. More recently, cross-institute affinity groups with common scientific themes have been established.

In August 2019, I explored the state of intramural NIH research with the longest-standing director of an NIH Institute, Dr. Anthony Fauci, who has directed the NIAID since 1984. I asked him whether the intramural program could justify receiving 10% of the overall NIH budget, which some researchers outside NIH have been known to question. Dr. Fauci agreed this allocation should always be up for discussion, and he understood it being an issue when funding levels for extramural research were so tight. He stated that intramural NIH research has some similarities to academic research but stressed there must be enough of a difference to add real value. Because IRP PIs do not spend time writing grant proposals, this should enable them, he said, to take on high-risk and big-impact research questions with long-term horizons.

The talent model for NIH faculty deserves mention. Freedom from grants in a large and diverse scientific environment is attractive, but limits on salary and ability to engage in entrepreneurial activity, along with the impediments of government bureaucracy, deter many from choosing to work there.

Institutes, laboratories, branches, and sections at NIH undergo periodic scientific review, and Fauci assured me those at his institute are quite rigorous. However, despite its very large budget and research carried out on a wide array of topics, the IRP is not generally viewed today as a hotbed of risky, long-term, large-scale, paradigm-shifting research. Given its size, scope, and organizational structure, whether this is indeed the case, and if so, why, should receive increased attention.

Overview of the Institutional Ecosystem

In the foregoing discussion, I have sought to present a novel taxonomy of institutions at which biomedical research is conducted in the US today. The ecosystem in which these institutions exist arose over decades in response to scientific, sociologic, and financial incentives, and it continues to evolve. The hospitals, medical schools, research institutes, and NIH intramural programs discussed here do not fully represent the extensive diversity of several thousand such institutions, but they may be sufficiently informative to permit a useful initial analysis that will stimulate subsequent inquiries and perspectives. Whether factors such as size of research programs, their geographic localization, their association with institutions considered "elite" or not, and many other factors are important considerations for this taxonomy and its implications will have to await future investigation.

I believe the success of the current ecosystem in part reflects this institutional diversity. Although the goal of the current review is not to identify winners and losers, it's important to take note of some of the attributes—both positive, negative, and of uncertain impact—of these institutional types. Perhaps surprisingly, there is scant published literature assessing the productivity of investments in these diverse settings, and little consensus on how such research might be conducted or evaluated. Why this is the case is a question for the sociology of science, but one consequence is our limited insight at present into what mix of institutions might maximize the quality and impact of future biomedical research, and which current institutional practices might be incentivized to undergo beneficial change. Key questions might include how these institutions set research goals and go about employing, funding, and enabling the success of their faculty.

Research Goals

The research goals of these institutional types both overlap substantially and differ in important ways. One major axis of distinction is between "basic science" research that seeks fundamental insights into biological processes, and research seeking to employ such insights (and others) to prevent and treat disease. It has been suggested that overly rigid distinctions between basic and "translational" research have become less useful or even anachronistic, though these terms do continue to be widely used (Flier and Loscalzo 2017). Unsurprisingly, research conducted in hospitals (whether associated with schools/universities or not) focuses on human health across a range of topics, from fundamental studies to more prevalent clinical/translational research on the diseases that hospitals and their faculties directly confront. In contrast, medical schools that function independent of hospitals (and universities without medical schools or hospitals) emphasize fundamental mechanisms, rather than disease-oriented therapeutic research, though the latter does also occur. Disease-focused research institutes investigate both basic and clinical/translational research relevant to their chosen disease. Research institutes that are not disease focused typically pursue fundamental research in any area of bioscience, believing such research will eventually uncover disease mechanisms and novel therapeutics that might not emerge from more disease-focused efforts. Whether research is basic or translational/clinical, these institutions all seek to patent and license discoveries that might have practical potential. Finally, the NIH IRP portfolio includes basic, translational, and clinical research across a broad array of areas.

Research Goals of Individual Faculty

The research goals of individual faculty in hospitals, medical schools, and disease-focused institutes tend to align with the interests and goals of the departments/divisions recruiting them, though this may change over time in response to scientific opportunities. At hospitals, the major organizational principle involves clinical specialties, while at medical schools and universities research is typically oriented around discipline-directed basic science departments. These organizational principles influence faculty research in complex ways. At research institutes that are not disease focused, faculty are often encouraged to pursue whatever ideas/hypotheses excite them independent of diseases or disciplines, their success later judged by the importance of their discoveries. Faculty at NIH IRP pursue research according to agreements between institute leadership and faculty, linked to the research goals of each hiring unit.

Recruitment of faculty is a critical requirement for institutional success, and approaches vary across these institutional categories, for both cultural and financial reasons. In hospital-based research, most faculty ascend from trainee positions (as MD and or PhD investigators) in diverse clinical areas, with a much smaller number identified via open searches. This reflects the diverse talent pool hospitals require and attract, and the institutional desire to build research expertise in multiple clinical domains. It is also a less expensive way to build a large faculty. In contrast, medical schools and universities with basic research programs recruit the great majority of their much smaller faculty into tenure-track positions via open searches, seeking to fill perceived scientific gaps or needs for science education. Disease-focused research institutes recruit faculty similarly to academic hospitals, though there may be a higher proportion of open external searches. Research institutes with a broad focus generally conduct open searches. Finally, the NIH IRP identifies faculty by open searches conducted by each institute/branch, recruited into positions that fulfill the goals of those units under government employment guidelines.

Funding Faculty Research

Approaches to funding faculty research also vary across these institutional categories. Upon hiring, hospitals, medical schools, and research institutes provide highly variable levels of institutional support (from nothing to millions of dollars based on institutional capacity and local traditions). Following this initial committed support (sometimes later supplemented if available by additional institutional funds), ongoing research almost always requires successful competition for external funding. Predominant external sources are the NIH, HHMI in a small minority, and a variable mix of foundation grants, corporate-sponsored research, and philanthropy. In stark contrast, institutes like Janelia and Arc fully support faculty salary and research operations with institutional funds; external grant support is neither needed nor permitted. This is also the case for the NIH IRP, whose budget derives from the federal government.

How institutions set their overall research goals and approach hiring, funding, and managing the careers of their faculty creates institutional cultures that must influence the efficiency, quality, and impact of their research. We unfortunately lack objective research capable of demonstrating and evaluating the connections between institutional practices and research outcomes. Comparative institutional efficacy is poorly understood at present and should be the subject of future research.

Implications and Paths Forward

This proposed taxonomy and the associated descriptive account of selected US biomedical research institutions is anchored in the recognition that these institutions have, collectively, been remarkably successful. The account is motivated by my belief that future efforts to enhance the system require that we understand the distinct features of the current institutional ecosystem (Alberts et al. 2014).

Several points are worth noting. First, the current success of the US biomedical research ecosystem did not result from centralized planning of these institutional homes (MacLeod and Urquiola 2021). Rather, success was enabled by the ability of existing and newly created institutions to respond nimbly to opportunities and incentives—a capacity far less evident in the more centrally controlled European institutions that dominated research in the 19th century. But unsurprisingly, some of these same incentives may have later produced unintended negative consequences, many flowing from dependence on current approaches to NIH funding and soft money faculty support, which many believe promote overbuilding and short-term thinking, among other problems (Alberts et al. 2014). Once identified, such flaws should be judiciously corrected, taking care to avoid disrupting the success of the ecosystem in the process. Several suggestions for future efforts follow.

Conduct More Research About Research

We should encourage more research into how we organize, fund, and evaluate biomedical research (Azoulay 2012), the value of which should be obvious to a community of scientists. Indeed, this topic is increasingly discussed today, and new and promising fields dubbed "meta-science," "research on research," and "the science of science" (SciSci) have emerged (Fortunato et al. 2018; Ioannidis et al. 2015; Wang 2021). So far, this field has produced many valuable insights and calls for action, but there have been few insights into comparative institutional effectiveness, the subject of this essay. At least two contributing factors seem likely. First, available tools to judge the quality, efficiency and impact of research on a large scale are imperfect (Hicks et al. 2015; Ioannidis and Khoury 2014). Surrogate indicators linked to publications, citations, patents, promotions and awards, acquisition of competitive funding, and clinical impact exist and are useful, but each has limitations, and collectively they need refinement. Research on improved metrics is needed (Hatch and Curry 2020). Second, most institutions lack incentives to objectively judge the efficiency and impact of their own research, perhaps fearing they might compare unfavorably to others. Instead, they focus on declaring their success in order to both please their faculties and justify a greater allocation of available resources from funding agencies and philanthropists.

Critically Evaluate NIH Funding Approaches

Since the existing ecosystem largely evolved through incentives created by funding mechanisms, and since the NIH is the dominant funder of biomedical research, additional attention must be focused on how NIH funding mechanisms—both intramural and extramural—produce beneficial or adverse outcomes (Azoulay and Li 2020). A systematic examination of funding mechanisms is beyond the scope of this paper, but the topic requires brief consideration, since NIH funding has shaped how most institutions came to be the way they are, which is the current topic.

Several major questions about NIH funding and brief preliminary answers follow.

The first question is what should be the overall size and growth rate of the NIH budget. Given its central role in the conduct of biomedical research in the US, there are few more important (and more politically complex) questions than this. A healthy research community requires the overall NIH budget to grow at a predictable and sustainable rate, designed to maintain—and ideally gradually increase—buying power. Major accelerations (the "doubling" between 1998 and 2003) and decelerations (including the inevitable deceleration after the unsustainable doubling) are harmful and should be avoided. A well-defined strategy to achieve this goal should be developed.

The second question is how might NIH funding (both direct and indirect) be redesigned to incentivize a more productive research environment. Efforts to shorten the excessively long time from receipt of proposals to their review and then funding should be readily achievable and would likely boost the effectiveness of the system. Additionally, funding mechanisms should be redesigned to incentivize gradual transition from largely soft-money faculty salary compensation environments—which likely adversely affect the quality of the workforce and its research output—to one where institutions are expected to use internal funds to support significant elements of their faculty's research time and effort. Given the diversity of institutional arrangements and wealth, the trade-offs required to accomplish this require substantial discussion and analysis. Further, we should reassess details of indirect (IDC) funding mechanisms that in their current form likely stimulate overbuilding of facilities and excessive growth of faculties who are given "hunting licenses" to populate newly built facilities with funded research. In the end, we may need a somewhat smaller population of better-supported researchers, an idea resisted by many leaders of the community, for whom growth is always assumed to be desirable.

Third, we need to ask how the NIH peer review and funding mechanisms might be redesigned to promote greater innovation and risk-taking (rather than incrementalism), longer-term project horizons, and increased collaboration and data-sharing. While it's unlikely that "evaluating people, not projects" as employed by HHMI can (or should) be scaled to the much larger community, approaches to rebalancing funding towards people versus projects should be considered. Funding decisions should be designed to better reflect judgments about scientists' past contributions and potential for innovative discoveries, rather than basing them excessively on preliminary data and proposals that claim to predict specific results. The possibility of failure is unavoidable when pursuing innovative discoveries, and current funding approaches diminish risk-taking and the ability to undertake projects with longer time horizons. Adding a cohort of successful but retired scientists to review panels might encourage reviews that seek reasons to fund interesting proposals, rather than reasons not to fund them, as may be more typical from young reviewers aiming to demonstrate their critical skills.

The approach to NIH funding is difficult to modify, enmeshed as it is in the political system, subject to special interest lobbying and stasis from political gridlock. While efforts to modify NIH approaches must continue, completely new funding mechanisms and institutional venues must also be encouraged and assessed.

Further Engage Committed Philanthropists

Many wealthy philanthropists have taken a major interest in bioscience research and have been willing to invest in new institutional models, including Rockefeller, Broad, HHMI/Janelia, and Arc discussed here. Other philanthropists (Gates, Parker, Bezos, Wyss, Chan-Zuckerberg, and others) with substantial capacity and interest have chosen to fund research at existing institutions rather than new ones, exerting influence through their ability to suggest topics, scientific approaches, and investigators. Though objective research is lacking, I suspect these philanthropically enabled entities and activities produce discoveries beyond their proportionate share of research expenditures. Whether this is true should be objectively studied, to identify practices that might be more broadly applied.

So, one important path to improving the current bioscience ecosystem would be to identify actionable insights into comparative institutional effectiveness. This would involve identifying connections between funding mechanisms and institutional effectiveness and whether philanthropically funded institutional arrangements might provide models for broader adoption.

Conclusion

I have sought here to provide a new taxonomy of the not-for-profit institutions in which biomedical research is conducted in the US today, reflecting differences in institutional missions, governance, faculty, funding, and culture. Consistent with limited prior scholarship, I conclude this diversity—born of decentralization and substantial freedom to operate—has contributed to the success of the nation's biomedical research ecosystem. As with financial investment strategies, a portfolio approach to research institutions is likely beneficial. But as in finance, frequent assessments and portfolio modifications are required.

The overarching goal of biomedical research is to elucidate the biology of human diseases and then to apply this knowledge to enhance human health. An effective portfolio of institutions will likely vary in several dimensions, including the research questions on which they focus, the ratio of basic versus applied initiatives, the funding mechanisms they employ, the timeframes and risk tolerance of typical projects, their approach to recruiting and supporting faculty and research infrastructure, and their approach to scientific and institutional leadership.

Much remains to be learned before we can attempt to optimize these complex and interacting domains. I hope this essay stimulates further research into the characteristics of our nation's biomedical research institutions, including those not discussed here from which important lessons can be learned. The motivation for this effort should be clear: despite substantial success, the US biomedical research ecosystem can surely be made even more effective.