Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 153
APPENDIX D THE IMPACT OF PUBLICLY FUNDED BIOMEDICAL AND HEALTH RESEARCH: A REVIEW1 Bhaven N. Sampat Department of Health Policy and Management Columbia University I. INTRODUCTION AND BACKGROUND New biomedical technologies trigger a number of major challenges and opportunities in health policy. Among economists, there is widespread consensus that new technologies are the major drivers of increased healthcare costs but at the same time a major source of health and welfare improvements (Murphy and Topel 2003). This has led to discussion about whether technological change in medicine is “worth it” (Cutler and McClellan 2001). The impact of new technologies on the health care system has also been the subject of much debate among health policy scholars more generally (Callahan 2009). Public sector research agencies have an important role in the U.S. biomedical innovation system. In 2004, federal agencies funded roughly one-third of all U.S. biomedical R and D (Moses et al. 2005). The National Institutes of Health (NIH) accounted for three-quarters of this amount. Private sector drug, biotechnology, and medical device companies provide the majority of U.S. biomedical R and D funding (about 58 percent). This private sector research is, in general, focused more downstream and tends to be closer to commercial application than NIH-funded research. 1 I thank Pierre Azoulay, and participants in the National Academies’ 2011 Workshop on Measuring the Impacts of Federal Investments in Research, for useful comments and suggestions. 153
OCR for page 154
154 MEASURING THE IMPACTS OF FEDERAL INVESTMENTS IN RESEARCH Donald Stokes (1997) observes that the public values science “not for what it is but what it is for.” A perennial question in U.S. science and technology policy is what benefits taxpayers obtain from publicly funded biomedical research. Recent concerns about the clinical and economic returns to NIH funding in the post-doubling era reflect this emphasis. In this paper, we review the evidence on the effects of publicly funded biomedical research. Reflecting Stokes’s observation above, the review will focus on the health and economic effects of public research, rather than measures of scientific outcomes. Given the prominence of the NIH in funding this research, many of the published articles and research focus on this agency. The evidence examined includes quantitative analyses, and qualitative case studies, published by scholars from a range of fields. While we have made efforts to be broad, the references discussed should be viewed as representative rather than exhaustive. This review takes stock of the empirical methodologies employed and the types of data used; it also highlights common research and evaluation challenges, and emphasizes where existing evidence is more, or less, robust. We proceed as follows. In Section II, below, we discuss a stylized model of how public research funding affects health, economic, and intermediate outcomes. As Kline and Rosenberg (1986), Gelijns and Rosenberg (1994), and others have emphasized, the research process cannot be reduced to a neat, linear model. While we recognize this fact (and highlight it in our literature review) the simple model is still useful in helping to organize our discussion of theory and data on the effects of publicly funded research. In Section III, we discuss the empirical evidence. In Section IV, we discuss common evaluation difficulties. In Section V, we conclude. The empirical approaches, data sources, and findings of many of the studies reviewed are also summarized in Tables D1–D3. II. PUBLIC SECTOR RESEARCH AND OUTCOMES: AN OVERVIEW Figure D-1 is a simple model illustrating how the literature has conceptualized the health and economic effects of publicly funded biomedical research (and publicly funded research more generally):
OCR for page 155
155 APPENDIX D Fundamental New Drugs, Knowledge Devices 1 Instruments and Techniques Health Publicly Private Sector Outcomes Funded R&D R&D 2 (Costs?) Prototype 3 Drugs, Devices Improved Clinical Practice Applied Knowledge Better Health Behaviors FIGURE D-1 Publicly Funded R and D and Outcomes, Logic Model SOURCE: Sampat, 2011 The top arm of the model illustrates one important relationship: publicly funded R and D yields fundamental knowledge, which then improves the R and D efficiency of private sector firms, yielding new technologies (drugs and devices) that improve health outcomes.2 This conceptualization has been the essential raison-d’etre for the public funding of science since Vannevar Bush’s celebrated postwar report, Science, The Endless Frontier. For example, Bush asserted in 1945 that “discovery of new therapeutic agents and methods usually results from basic studies in medicine and the underlying sciences” (Bush 1945). It is also the essential mechanism in several important economic models of R and D (e.g. Nelson 1984). Importantly, this conceptualization generally views publicly funded research as “basic” research that is not oriented at particular goals, and thus yields benefits across fields. The influential “market failure” argument for public funding of basic research is that profit-maximizing, private-sector firms will tend to underinvest in this type of fundamental, curiosity driven research, since they cannot appropriate its benefits fully (Nelson 1959, Arrow 1962). The channels through which publicly funded basic research might influence private sector innovation are diverse, including dissemination via publications, presentations and conferences, as well as through informal networks (Cohen et al. 2002). Labor markets are another 2 Stokes (1997) and others have challenged this definition of “basic” research.
OCR for page 156
156 MEASURING THE IMPACTS OF FEDERAL INVESTMENTS IN RESEARCH channel, since public agencies may also be important in training doctoral and post-doctoral students who move on to work for private sector firms (Scherer 2000). The second arrow illustrates another relationship. New instruments and techniques that are by-products of "basic" research can also improve private sector R and D (Rosenberg 2000). Prominent examples of instruments and research tools emanating from academic research include the scanning electron microscope, the computer, and the Cohen- Boyer recombinant DNA technique. Third, publicly-funded researchers sometimes develop prototypes for new products and processes. Some of these are indistinguishable from the informational outputs of basic research discussed above. For example, when academic researchers learned that specific prostaglandins can help reduce intraocular pressure this discovery immediately suggested a drug candidate based on those prostaglandins, though the candidate required significant additional testing and development. (This academic discovery later became the blockbuster glaucoma drug, Xalatan.) The public sector has also been important in developing prototypes (Gelijns and Rosenberg 1995). Roughly since the passage of the Bayh-Dole Act, in1980, publicly funded researchers have become more active in taking out patents on these inventions and prototypes for new products and processes, and licensing them to private firms (Mowery et al. 2004. Azoulay et al. 2007). While much of the discussion of publicly funded biomedical research focuses on this more “basic” or fundamental research the public sector also funds more “applied” research and development.3 For example, about one-third of the NIH budget is for clinical research, including patient oriented research, clinical trials, epidemiological and behavioral studies, as well as outcomes and health services research. Such research can be a useful input into the development of prototypes, and may also directly inform private sector R and D. Clinical research may also directly affect health behaviors. For example, knowledge from epidemiological research about cardiovascular health risk factors contributed to reductions in smoking and better diets (Cutler and Kadiyala 2003). New applied knowledge can also influence physicians: 3 Stokes (1997) provides a thoughtful critique of conventional distinctions between “basic” and “applied” research. Since much of the literature before and since Stokes uses this terminology, we employ it in our review of this literature, even while recognizing the importance of his argument.
OCR for page 157
157 APPENDIX D for example, by changing their prescribing habits (e.g. “beta-blockers after heart attacks improve outcomes”) or routines (e.g. “this type of device works best in this type of patient”). Importantly, as various studies we review below will emphasize, negative results from clinical trials—showing that particular interventions do not work — can also be important for clinical practice and in shaping health behaviors. While the discussion above assumes that new biomedical knowledge and technologies improve health outcomes, this is a topic of debate. The conventional wisdom is that while other factors (e.g. better diet, nutrition, and economic factors) were more important for health outcomes historically (McKeown 1976), improvements in American health in the post-World War II era have been driven largely by new medical knowledge and technologies (Cutler, Deaton, and Lleras-Muney 2006). The contribution of publicly funded research to these developments is an open empirical question, discussed below. At the same time, some scholars suggest that we may have entered an era of diminishing returns, where new technologies are yielding increasingly less value (Callahan 2009; Deyo and Patrick 2004). The effect of new biomedical technologies on healthcare costs is a related concern. There is general agreement among health economists that new medical technologies are the single biggest contributor to the increase in long-run health costs, accounting for roughly half of cost growth (Newhouse 1992). Rising health costs strain the budgets of public and private insurers as well as employers, and may also contribute to generate health inequalities. The dynamic that exists between new medical technologies and health costs in the U.S. may reflect a "technological imperative," which creates strong incentives for the healthcare system to adopt new technologies once they exist (Fuchs 1995; Cutler 1995). It may also reflect positive feedbacks between demand for insurance and incentives for innovation (Weisbrod 1991). Concern about the effects of technology on health costs has fueled empirical work on whether technological change in medicine is "worth it." Long ago, Mushkin (1979) noted (though did not share) “widespread doubt about the worth of biomedical research given the cost impacts.” A large literature in health economics suggests that new biomedical technologies are indeed, in the aggregate, worth it. Cutler (1995) and others suggest that, given the high value of improved health (current estimates suggest the value of one additional life year is $100,000 or
OCR for page 158
158 MEASURING THE IMPACTS OF FEDERAL INVESTMENTS IN RESEARCH more), even very costly medical technologies pass the cost-benefit test.4 Nordhaus (2003) estimates that the value of improvements in health over the past half century are equal in the magnitude to measured improvements in all non-health sectors combined. Others (Callahan 2009) view these health cost increases as unaffordable, even if they deliver significant value, and therefore ultimately unsustainable. At the same time, not all medical technologies necessarily increase costs. As Cutler (1995) and Weisbrod (1991) indicate, technologies that make a disease treatable but do not cure it - moving from non-treatment to "halfway" technology in Lewis Thomas's characterization-are likely to increase costs. The iron-lung to treat polio is an example of this. However, technologies that make possible prevention or cure ("high technology") can be cost-reducing, especially relative to halfway technologies. Thus the polio vaccine was much cheaper than the iron lung. Consistent with this, Lichtenberg (2001) shows that while new drugs are more expensive than old drugs, they reduce other health expenditures (e.g. hospitalizations). Overall, he argues, they result in net decreases in health costs (and improve health outcomes).5 As Weisbrod (1991) notes, "The aggregate effect of technological change on health care costs will depend on the relative degree to which halfway technologies are replacing lower, less costly technologies, or are being replaced by new, higher technologies. " 6 One way to think about the effects of public sector spending on costs would be to assess the propensity of publicly funded research to generate (or facilitate the creation of) these different types of technologies. However, since the effects of these new technologies are mediated by various facets of the health care and delivery system, it may be difficult conceptually (and empirically) to isolate and measure the effects of public sector spending on overall health costs (Cutler 1995).7 4 Cutler (1998) observes "Common wisdom suggests that rapid cost increases are necessarily bad. This view, however, is incorrect. Cost increases are justified if things that they buy (increases in health) are worth the price paid." (2) 5 See however, Zhang and Sourmerai (2007) for a critique of this finding. 6 The cost-effectiveness of these technologies also depends on the populations on which they are used, as Chandra and Skinner (2011) emphasize. 7 There is also some discussion about whether the public sector should be paying attention to the cost-side consequences of its investment decisions. Weisbrod (1991) notes: "With respect to the NIH, it would be useful to learn more about the way the size and allocation of the scientific research budget are influenced,
OCR for page 159
159 APPENDIX D III. THE EFFECT OF PUBLICLY FUNDED RESEARCH: A REVIEW OF THE EVIDENCE Health Measuring the health returns to publicly funded medical research has been a topic of interest to policymakers for decades. In an early influential study, Comroe and Dripps (1976) consider what types of research (basic or clinical) are more important to the advance of clinical practice and health. The authors rely on interviews and expert opinion to determine the top ten clinical advances in the cardiovascular and pulmonary arena, and identified 529 key articles associated with these advances. They coded each of the key articles into six categories: (1) Basic research unrelated to clinical problems; (2) Basic research related to clinical problems (what Stokes later termed “use-oriented” basic research); (3) Research not aimed at understanding of basic biological mechanisms; (4) Reviews or syntheses; (5) Development of techniques or apparatuses for research; and (6) Development of techniques or apparatuses for clinical use. The authors find that 40 percent of the articles were in category 1, and 62 percent in categories 1 or 2. Based on this, the authors assert “a generous portion of the nation's biomedical research dollars should be used to identify and then to provide long-term support for creative scientists whose main goal is to learn how living organisms function, without regard to the immediate relation of their research to specific human diseases.” Comroe and Dripps also note “that basic research, as we have defined it, pays off in terms of key discoveries almost twice as handsomely as other types of research and development combined” (1976). A more recent set of studies examines the effects of publicly funded research on health outcomes. Operationalizing the concept of “health” is notoriously difficult. Common measures employed to account for both the morbidity and mortality effects of disease include quality adjusted life years (QALYs) and disability adjusted life years (DALYs) (Gold et al, 2002). However, it is difficult to get longitudinal information on these measures by disease. As a result, most of the analyses of the effects of public funding on health examine more blunt outcomes, including the number of deaths and mortality rates for particular diseases. perhaps quite indirectly, by the health insurance system, through its impact on the eventual market for new technologies of various types" (535).
OCR for page 160
160 MEASURING THE IMPACTS OF FEDERAL INVESTMENTS IN RESEARCH Numerous prominent academic studies (Weisbrod 1983, Mushkin 1979) aim to examine the health effects of biomedical research, and the economic value of this impact, in a cost-benefit framework. One important recent study in this tradition, Cutler and Kadiyala (2003), focuses on cardiovascular disease—the disease area where there has been the strongest improvement in health outcomes over the past sixty years. Since 1950 mortality from cardiovascular disease decreased by two- thirds, as Figure D-2 (reprinted from their paper) shows: 450 400 350 Deaths per 100,000 300 250 200 150 100 50 0 19 0 19 2 19 4 56 19 9 61 19 3 19 5 19 7 68 19 0 72 19 4 19 6 78 19 9 19 1 19 3 19 5 87 19 9 19 1 93 5 5 5 5 6 6 6 7 7 7 7 8 8 8 8 9 19 19 19 19 19 19 19 CVD Cancer Infections HIV Accidents, Homicide, Suicide Other FIGURE D-2 Mortality by cause of death 1950-1994 SOURCE: Cutler and Kadiyala 2003 Cutler and Kadiyala, through a detailed review of the causes of this advance (relying on epidemiological and clinical data, medical textbooks, and other sources), estimate that roughly one third of this cardiovascular improvement is due to high-tech treatments, one third to low tech treatments, and one third to behavioral changes. Assuming one additional life year gained is valued at $100,000, the authors compute a rate of return of 4-to-1 for investments in treatments and 30-to-1 for investments in behavioral changes. These investments include costs borne by consumers and insurers, and estimates of public sector R and D for cardiovascular disease. Based on these figures, the authors argue that the rate of return to public funding is high, though they don’t directly trace public funding to changes in outcomes in their quantitative analyses. Interestingly, in their qualitative account, the major public sector research activities highlighted have an “applied” orientation, including the NIH’s role in
OCR for page 161
161 APPENDIX D sponsoring large epidemiological trials and holding consensus conferences. This may reflect a traceability and attribution problem, which is common to the evaluation of fundamental research: It is difficult to directly link improvements in outcome indicators to public sector investments in basic research, even in a study as detailed as this one. A paper by Heidenreich and McClellan (2003) is similarly ambitious, looking at sources of advance in the treatment of heart attacks. The authors focus on this treatment area, not only because of the large improvements, but also because it is a "best case" for attributing health outcomes to particular biomedical investments. Specifically, these authors go further than Cutler and Kadiyala by attempting to link changes in clinical practice to changes in specific R and D inputs. The authors focus here on clinical trials, not basic research. This is not because they believe that basic research is unimportant, “but because it is much easier to identify connections between these applied studies and changes in medical care and health.” Based on detailed analyses of MEDLINE-listed trials and health outcomes, the authors argue that medical treatments studied in these trials account for the bulk of improvement in AMI outcomes. The authors associate changes in clinical practice and outcomes to research results reported in trials through analysis of timing of events, and detailed clinical knowledge of how the trial results, clinical practices, and health outcomes relate. One interesting result from this paper is that clinical practice often “leads" formal trials, challenging the “linear” model embodied in Figure D-1 (above). The authors also emphasize that an important role for trials is negative: telling clinicians what doesn't work, and stopping the diffusion of ineffective technologies. While the sample they examine represents a mix of publicly funded and privately funded trials, the authors do emphasize a particularly important role for the public sector in funding trials on drugs off patent, where private firms have fewer incentives to do so. Philipson and Jena’s (2005) study of HIV-AIDS drugs is another paper that examines the value of increases in health from new medical technologies. Though this study does not explicitly focus on the role of the public sector, it estimates that HIV-AIDS drugs introduced in the 1990s generated a social value of $1.4 trillion, based on the value of the increments to life expectancy created from these drugs (here again, using the estimate of $100,000 per life year). This study is relevant because of
OCR for page 162
162 MEASURING THE IMPACTS OF FEDERAL INVESTMENTS IN RESEARCH the important role of public sector research in the development of HIV- AIDS drugs, which is observed in several of the empirical studies discussed below. A recent paper by Lakdawalla et al (2011) employs a similar approach to assess the benefits from cancer research. The authors find these benefits to be large, estimating the social value of improvements from improvements in life expectancy during the 1988-2000 period to be nearly $2 trillion. The authors note that this compares to investments of about $80 billion dollars in total funding for the National Cancer Institute between 1971 and 2000. As with the HIV studies discussed above, the authors do not calculate a rate-of-return on publicly funded research explicitly, but do argue that the social benefits from cancer research in general far exceed research investments and treatment costs. A large share of the benefits in the cancer arena, according to this work, results from better treatments. Lichtenberg (2004) also suggests that new drug development has been extremely important in progress against cancer.8 Public sector research may have been important to the development of these drugs: various studies (Stevens et al. 2011, Chabner and Shoemaker 1989) suggest an important role for the public sector in cancer drug development.9 Each of the studies discussed so far focuses on particular disease areas. In a more "macro" approach Manton (2009) and colleagues relate mortality rates in four disease areas to lagged NIH funding by the relevant Institute, over the period 1950-2004. They find that for two of the four diseases (heart disease, stroke) there is a strong negative 8 Cutler (2008) also emphasizes progress in the “war on cancer” – though highlights the role of screening and personal behavior changes, and notes the high costs of treatment. Sporn (2006) and Balilar and Gonik (1997) offer less sanguine assessments, emphasizing that progress against cancer has been highly uneven. Long-standing debates in assessments of the War on Cancer include the disagreements on the relative importance of treatment versus prevention, and of basic versus applied research. The literature also suggests it is difficult to evaluate the extent of progress in cancer, for two main reasons. First, advances in screening increase incidence. The second is competing risks: for example, the reduction in mortality from cardiovascular disease, discussed above, increased cancer cases. See Cutler (2008) for a review. 9 A National Cancer Institute (NCI) “Fact Sheet” asserts that “approximately one half of the chemotherapeutic drugs currently used by oncologists for cancer treatment were discovered and/or developed at NCI.” http://www.cancer.gov/cancertopics/factsheet/NCI/drugdiscovery
OCR for page 163
163 APPENDIX D correlation, but find weaker evidence for cancer and diabetes. Several issues arise here that will re-emerge in other quantitative analyses discussed below. First, linking funds to disease areas is difficult. As with other studies we will consider below, the authors here rely on the disease foci of Institutes within the NIH. More importantly, the counterfactual is hard to prove: It is difficult to make the case that the relationships estimated are causal, since Institute-specific funding is not exogenous. In particular, diseases where there is highest expectation of progress (even absent funding) may be more likely to get funds. Finally, competing risks also complicate interpretation of health outcomes. For example, part of the reason cancer mortality has increased rather than decreased over the period studied is that people no longer die of heart attacks, due to advances in the cardiovascular arena. Private Sector R and D Another set of studies relates publicly funded research to private sector R and D and productivity. These include econometric analyses relating public sector and private sector funding, surveys of firm R and D managers, and studies examining the geographic dimension of spillovers from public sector researchers. Several papers relate NIH funding by disease area to later private sector funding. One motivation in these studies is to assess if public and private sector R and D are substitutes or complements, an issue of perennial interest in science and technology policy (David, Hall, and Toole 2000). The econometric analyses generally find a positive association between public sector and private sector funding. Toole (2007) uses data from the NIH’S Computerized Retrieval of Information on Scientific Projects (CRISP) database, covering NIH basic and clinical research funding across seven therapeutic classes (between 1972 and 1996), and data from the Pharmaceutical Manufacturers of America (PhRMA) on private sector R and D in these same areas (between 1980 and 1999) to examine the relationships between the two. This study finds a 1 percent increase in basic research funding associated with a 1.7 percent increase in private sector funding, though the elasticity for clinical research is much smaller (.4 percent). In a similar analysis, Ward and Dranove (1995), using PhRMA data on R and D spending and NIH data on funding by Institute (similar to that used in the Manton et al 2009 study discussed above) find that a 1 percent increase in NIH research support in a disease area is associated with a .76 percent increase in
OCR for page 182
182 Authors Question Empirical Approach Measures/Data Results Comroe and What types of research Interviews, expert Top 10 clinical advances 41 percent of all work Dripps (clinical vs. basic) are opinions used to “Key articles” associated judged to be essential or (1976) important in the advance of determine of top 10 with these advances crucial for later clinical clinical practice, health? clinical advances in Coding of whether the advances was not cardiovascular and key articles are clinical clinically oriented at the pulmonary arena or non-clinical time of research Content analyses of key articles
OCR for page 183
TABLE D-2 Public Funding and New Drugs, Devices: Summary of Selected Studies Authors Question Empirical Approach Measures Results Cockburn and How does public Case studies of 15 clinically Qualitative Of 15 drugs, public sector Henderson sector research affect important drugs determinations of roles research made key enabling (1996) pharmaceutical of public sector in drug discovery for 11 innovation? development Public sector involved in synthesis of major compound in 2 cases Ward and How do MEDLINE Panel regressions articles in NIH data on R and D Strong relationship between Dranove “drug” articles a disease area to NIH R and by institute NIH funding and later (1995) respond to NIH D by relevant institute MEDLINE data on MEDLINE articles funding? publications by disease Indirect effect (from research area outside disease area) stronger than direct effect Sampat and What are the roles of Examine share of new FDA approved NMEs Direct effect: public sector Lichtenberg the public and private molecular entities where 1988-2005 owns key patent for 9% of (2011) sectors in drug public sector developed Orange Book patents drugs development? patent (direct effect) and on these drugs Indirect effect: Public sector where private sector patents Government interest patents or publications cited cite public sector statements/assignment by 48% of drugs patents/publications in patents Both direct and indirect (indirect effect) Backward citations in effects more pronounced for patents to public sector most clinically important patents, MEDLINE drugs (17%, 65%) 183
OCR for page 184
184 Authors Question Empirical Approach Measures Results articles acknowledging public sector funding Sampat (2007) On how many drugs Examine share of drug FDA approved NDAs 72 of 1546 NDAs have an do academic approvals where academic 1988-2005 academic patent institutions own and public sector Orange Book patents 10.3 percent of NMEs patents? institutions own key patents on these drugs 5.9 percent of non-NMEs USPTO data on patent 19.2 percent of priority NMEs ownership have an academic patent Azoulay-Sampat concordance of academic assignees Keyhani et al Do drug prices Regression analyses relating 180 drugs listed in the Government supported (2005) reflect development drug prices to measures of Federal Register clinical trials for 6.6 percent time and government government support between 1992 and 2002 of the drugs investment? Federal Register data Government owned or on their patents supported patents for 7.2 Information on percent of the drugs government assignees and government interest statements for these patents Data from NIH clinical trials database and FDA on whether NIH trials supported FDA approval
OCR for page 185
Authors Question Empirical Approach Measures Results Stevens et al On how many drugs Examine number of drug FDA data on drug and 153 FDA-approved drugs (2011) and vaccines approvals in-licensed from biologic approvals discovered by public sector emanate from public PSRIs (excluding licenses Orange Book data on institutions over past 40 years sector research to platform technologies) FDA approved drugs (102 NMEs, 36 biologics, 15 institutions? AUTM data on vaccines) academic patents and 13 percent of NMEs (21 licenses percent of priority NMEs) rDNA data on licensing licensed from public sector transactions research Virtually all important vaccines introduced over past 25 years come from public sector Broad correlation between NIH Institute budgets and therapy classes with public sector drugs Kneller (2010) How important are Examine place of 252 FDA approved Overall 24% of drugs from new employment of inventors on drugs 1998-2007 universities companies/universiti key patents for drugs Data on patents from By novelty: 31% of most es (and other actors) Orange Book, Merck scientifically novel drugs in drug discovery? Index, other sources By priority: 30% of priority- Data from concurrent review drugs publications and from interviews on inventors’ places of 185
OCR for page 186
186 Authors Question Empirical Approach Measures Results employment Morlacchi and What were the Longitudinal case study of Interview data NHLBI contracts important in Nelson (2011) sources of innovation the development of the Information from key spurring firm formation and behind development LVAD patents and evolution in 1960s/1970s of the left-ventricular publications on LVAD NHLBI important in assist device sponsoring conferences, (LVAD)? How centers to promote diffusion important was the of best practice among NIH? academics and industry Public funding of key trials and development of component technologies also important Application led scientific understanding; basic understanding of heart failure remains weak
OCR for page 187
Dorsey et al Are new drug Correlations of NIH 1995-2000 FDA drug Despite a rise in NIH (and (2009) approvals by funding data with future and approvals, mapped other funding), drug approvals therapeutic area drug approvals to nine disease areas flat overall associated with NIH NIH funding by Within class analyses of drug funding in those Institute; allocated to approvals also show little areas? disease areas based on correlation with research Congressional inputs justifications Note: Also estimate R and D by biotechnology firms, medical device firms, pharmaceutical companies, non-profits Blume-Kohut How does NIH Panel regression CRISP and Some evidence of (2009) funding in a disease RePORTER data on responsiveness of Phase I area relate to the NIH grants/funds 1975- trials: elasticity .25-.31 number of drugs 2004 No evidence of subsequently in Grants associated with responsiveness of Phase III Phase I and Phase III disease areas using trials trials in that area? parsing of abstracts, keywords, concordance with MeSH thesaurus PharmaProjects data on drugs in development, by phase and category 187
OCR for page 188
188 TABLE D-3 Public Funding and Private R and D, Patenting: Summary of Selected Studies Authors Question Empirical Approach Measures Results Ward and How does industry Panel regressions PhRMA data on R and A 1 percent increase in NIH Dranove funded R and D relating private R and D D by field research associated with .76 (1995) respond to NIH R in a disease area to NIH NIH data on R and D by percent increase by private sector and D? R and D by relevant institute over next seven years (direct) institute Controls for disease A 1 percent increase in NIH burden, drug research associated with 1.7 development, time percent increase by private sector over next seven years (indirect) Contemporaneous correlations highest Cockburn How does interaction Panel regression models MEDLINE data from Statistically significant association and with public sector relating productivity to35,000 articles on firms’ between propensity to co-author Henderson science within firm variation inco-authorship, with academics and important (1996) (collaboration, hiring interaction with public publication by “star” patents/dollar of “star” scientists) sector, with firm fixed scientists for 10 firms, Statistically significant association affect firm-level R effects 1980-1988 between share of publications and D productivity Data on “important” from “star” scientists and patents/R and D for important patents/R and D dollar these firms Toole (2007) Does public Panel regression models CRISP data on NIH Public and private sector research scientific research relating pharmaceutical basic and clinical complements complement private R and D by to NIH research mapped to 7 A 1 percent increase in basic
OCR for page 189
Authors Question Empirical Approach Measures Results R and D investment? funding across disease therapeutic classes, research funding associated with a areas, over time 1972-1996 1.7 percent increase in private PhRMA data on private sector R and D sector R and D in these A 1 percent increase in clinical classes, 1980-1999 research funding associated with a .40 percent increase in private sector R and D Azoulay, Do elite life scientists Panel regression models Data on 10,450 elite life Professional transitions lead to a Graff Zivin, benefit local firms? examining geography of science researchers decrease in citations (in patents Sampat citations to scientists’ (most publicly funded) and articles) to movers’ pre-move (2011) work before and after Historical information patents at original location they move on productivity, Weaker evidence of increase in employment locations of citations from firms at destination each scientist location MEDLINE data on their publications ISI data on citations to their publications USPTO data on their patents USPTO data citations to their patents and publications Zucker, How important was Panel regression models 337 “star” scientists Presence of stars and their Darby and academic science in relating location of new (based on articles, collaborators – “intellectual Brewer the creation of new biotechnology firms to genetic discoveries in capital” – in an area has a 189
OCR for page 190
190 Authors Question Empirical Approach Measures Results (1998) biotech firms? number of “star” Genbank) statistically significant and scientists in area Data on their positive relationship with the collaborators number of new biotechnology Location and affiliation enterprises later formed in that of stars (from journal area articles Data on biotechnology firms and firm formation form North Carolina Biotechnology Center and Bioscan Cohen, What are the roles of Survey 1994 Carnegie Mellon Pharmaceutical industry an Nelson, public sector Survey of Industrial R outlier: reports public research the Walsh research on industrial and D managers most important source of new (2002) R and D? What are Merged with publicly project ideas and contributing to the channels through available data on project completion which public respondents Medical instruments industry R research affect and D projects less frequently use industrial R and D? any of three outputs of public research than other industries Drug industry makes use of public research much more frequently Top three fields contributing to R and D in pharmaceuticals: Medicine, Biology, Chemistry Top three fields contributing to R
OCR for page 191
Authors Question Empirical Approach Measures Results and D in medical instruments industry: Medicine, Materials Science, Biology Mansfield How important is Survey Survey results from 77 Percent of new products that could (1998) academic work for firms not have been developed (without industrial substantial delay) in absence of innovation? recent academic research, 1986- 1994: 31 in drugs/medical products (15 across all industries) Percent of new processes that could not have been developed (without substantial delay) in absence of recent academic research, 1986-1994: 11 in drugs/medical products (11 across all industries) 191
OCR for page 192