Funding of US Biomedical Research, 2003-2008

Context With the exception of the American Recovery and Reinvestment Act, funding support for biomedical research in the United States has slowed after a decade of doubling. However, the extent and scope of slowing are largely unknown.

Objective To quantify funding of biomedical research in the United States from 2003 to 2008.

Design Publicly available data were used to quantify funding from government (federal, state, and local), private, and industry sources. Regression models were used to compare financial trends between 1994-2003 and 2003-2007. The numbers of new drug and device approvals by the US Food and Drug Administration over the same period were also evaluated.

Main Outcome Measures Funding and growth rates by source; numbers of US Food and Drug Administration approvals.

Results  Biomedical research funding increased from $75.5 billion in 2003 to $101.1 billion in 2007. In 2008, funding from the National Institutes of Health and industry totaled $88.8 billion. In 2007, funding from these sources, adjusted for inflation, was $90.2 billion. Adjusted for inflation, funding from 2003 to 2007 increased by 14%, for a compound annual growth rate of 3.4%. By comparison, funding from 1994 to 2003 increased at an annual rate of 7.8% (P < .001). In 2007, industry (58%) was the largest funder, followed by the federal government (33%). Modest increase in funding was not accompanied by an increase in approvals for drugs or devices. In 2007, the United States spent an estimated 4.5% of its total health expenditures on biomedical research and 0.1% on health services research.

Conclusion After a decade of doubling, the rate of increase in biomedical research funding slowed from 2003 to 2007, and after adjustment for inflation, the absolute level of funding from the National Institutes of Health and industry appears to have decreased by 2% in 2008.

Biomedical research is valued highly by individuals, governments, foundations, and corporations. Research is seen as a source of more effective treatments and preventive measures and as a route to political policy, economic development, and new commercial products.

In 2005, we reported that total public and private financial support of US biomedical research increased substantially during the preceding decade, with a tripling in nominal amount and doubling after adjustment for inflation between 1994 and 2003.¹ However, as measured by new drugs approved by the US Food and Drug Administration (FDA), productivity has been stagnant. The decrease in number of novel drugs (new molecular entities) approved was striking. Medical devices, especially implantable and diagnostic devices, fared better during this interval, as reflected in number of approvals, companies' financial performance, and increased investment in research and development.

The 2007-2009 severe global recession has renewed focus on public spending and has caused companies and foundations to reexamine their priorities. Also, debate in the United States about the role of the federal government in providing health insurance has cast attention on the allocation of research support, especially between discovery of new clinical interventions and evaluation of their effect, value, and cost. The US government is following Europe's lead in providing incentives for health services research and information technology, which were incorporated in the American Recovery and Reinvestment Act of 2009.²

Researchers and sponsors are increasingly aware that financing is necessary, but not sufficient, to sustain progress. While the promise of new drugs for refractory common or devastating diseases continues to capture the public's imagination and enjoys strong support, policy makers are also aware that new beneficial technology often spawns new cost. Consequently, timely and accurate information about the sources of public and private research funds is important.

Therefore, we updated our earlier analysis of 1994-2003 to 2008, where data were available. This reexamination allowed us to confirm some trends seen earlier, gauge the impact of the recession, and improve our previous estimates.

Our goal in the current study was to replicate and extend our 2005 study¹ by providing updated and improved data through 2007 for all sources and 2008 data where available. Generally, data sources were similar to those in our previous study, with the notable exception that we used different sources to identify research support from biotechnology and medical device firms to minimize double counting of support received from life sciences firms.¹ Four major sponsors of biomedical research were considered: federal government; state and local governments; private, not-for-profit entities including foundations; and industry. As in the previous study, biomedical research was defined as life sciences, excluding agricultural science but adding psychology. Data regarding subsidies from colleges and universities were updated. Publicly available information was used wherever possible.

As in the previous study, values were adjusted to 2008 dollars using the biomedical research and development price index (BRDPI).³ A preliminary value for 2008 BRDPI was used. The BRDPI “measures changes in the weighted-average of the prices of all the inputs (eg, personnel services, various supplies, and equipment) purchased with the [National Institutes of Health {NIH}] budget to support research” and in theory indicates how much NIH expenditures would need to increase to maintain research activity at a constant level.⁴

National Institutes of Health funding was taken from the National Science Foundation Division of Science Resources Statistics data on obligations for research, development and research and development plant, 2003-2008.⁵ The NIH defines obligation “based on NIH funds that have been awarded by an NIH Institute/Center.”⁶ Similar to our earlier study, federal funding was determined using data from the National Health Expenditure Accounts (NHEA),⁷ which has revised its method of data collection and updated all data from 1976 forward using data directly from the NIH instead of from the National Science Foundation.⁸ We also used NHEA data to determine state and local government support for biomedical research.

Private entities considered included foundations, public charities, medical research organizations (eg, Howard Hughes Medical Institute), and voluntary health organizations (eg, American Cancer Society). Funding was determined from these organizations using NHEA data.

Industry contribution to biomedical research comes from 3 major sources: pharmaceutical, biotechnology, and medical device firms. We used publicly available data, including those from trade organizations and firms' financial reports, to determine industry support. For pharmaceutical companies, data from the Pharmaceutical Research and Manufacturers of America (PhRMA) trade organization were used,⁹ and only domestic research and development expenditures by its member companies were considered. Our previous study was limited by double counting of research support from firms that were members of PhRMA and of Biotechnology Industry Organization. In this analysis, data were used from Burrill & Company on research and development expenditures of biopharmaceutical companies that were not members of PhRMA.^10-13 Such estimates were available for 2004-2008 but not for 2003. Therefore, linear regression from 2004-2008 was used to estimate funding for 2003.

Incremental funding from medical device firms was estimated by totaling research and development expenditures in the annual reports¹⁴ of the 20 largest (measured by 2008 revenues) US-based firms that were not members of PhRMA. In our previous study, we relied on an external report that has not been updated for support from medical device firms. Data from the FDA were also used to quantify the number of novel treatments, including small-molecule biologics¹⁵ and devices,¹⁶ that were approved for use between 2003-2008.

We estimated biomedical research funding at academic institutions from the National Science Foundation's Academic Research and Development Expenditures report.¹⁷

We derived health services research expenditures from 2003-2008 from the Coalition of Health Services Research annual reports,¹⁸ which includes health services research funding from the Agency for Healthcare Research and Quality, the Centers for Disease Control and Prevention, the Centers for Medicare & Medicaid Services, the NIH, and the Veterans Health Administration. Because updated reports on health policy grant making, which we used in our previous study, were not available, foundation support for health services research was estimated solely from the annual reports of the Robert Wood Johnson Foundation,¹⁹ which accounted for 63% of health policy funding by foundations in 2002.

We used polynomial regression models to assess financial trends, allowing for a quadratic shape over time. Model parameters were estimated separately for early (1994-2003) and later (2003-2007) periods and were used to estimate mean annual growth rates throughout each period. Mean rates were compared using t tests at a 2-sided significance level of 5%, with model-based standard error estimates obtained using the delta method.

A single imputation was used to estimate funding for biotechnology firms in 2003. A linear regression model was used to predict these missing data based on available data in this period (2004-2008). Analyses were conducted using SAS version 9.2 (SAS Institute Inc, Cary, North Carolina).

Total funding for biomedical research increased from $75.5 billion in 2003 to $101.1 billion in 2007 (Table 1 and Figure 1). Adjusted for inflation using the Biomedical Research and Development Price Index, this represents an increase of 14% over 4 years, from $92.3 billion in 2003 to $105.6 billion in 2007. By comparison, the gross domestic product (GDP) of the United States, adjusted for inflation, increased by 12% in the same period.²⁰ In our previous study, funding increased at a compound annual growth rate of 7.8% for 1994-2003 compared with a compound annual rate of 3.4% for 2003-2007 (P < .001). Funding data for 2008 were available only for the NIH and industry and totaled $88.8 billion. The corresponding total for 2007 for the NIH and industry was $86.4 billion and when adjusted to 2008 dollars is $90.2 billion, indicating that real (adjusted for inflation) funding for biomedical research from the NIH and industry decreased from $90.2 billion in 2007 to $88.8 billion in 2008.

As in the previous study, industry remained the largest contributor to biomedical research, accounting for 58% of all expenditures in 2007. The NIH remained the second-largest contributor, accounting for 27% of expenditures, followed by state and local governments (5%), non-NIH federal sources (5%), and private not-for-profit support (4%).

The NIH remains the largest federal contributor to biomedical research, accounting for 84% of total federal funding in 2007. Adjusted for inflation, NIH contributions decreased by 8.6%, from $31.8 billion in 2003 to $29.0 billion in 2007 (P < .001 for the funding trend from 2003-2007 compared with 1994-2003). When adjusted for inflation, total federal funding increased by $200 million (0.7%) from 2003-2007. In our previous study, total federal funding increased by nearly 100% from 1994-2003 (P < .001).

Because of methodological changes in data collection under the NHEA,⁸ the calculated federal expenditures, state and local funding, and private not-for-profit funding for the years 2003-2004 are modestly different from those in the previous study.¹

State and local government spending on biomedical research increased from $4.2 billion in 2003 to $5.2 billion in 2007, or 6% after adjusting for inflation. State and local government expenditures had increased 45% from 1994-2003 after adjustment for inflation (P < .001).

Foundations, public charities, medical research organizations, and voluntary health organizations contributed $4.3 billion in 2007, an increase of 11% from 2003 when adjusted for inflation (P = .42 for the funding trend from 2003-2007 compared with 1994-2003).

Support from pharmaceutical, biotechnology, and medical device companies increased from $40.0 billion in 2003 to $58.6 billion in 2007, an increase of 25% when adjusted for inflation. The largest contributor to biomedical research remained the pharmaceutical sector, followed by biotechnology firms and then medical device companies. Conversely, medical device firms demonstrated the largest rate of growth (59%, adjusted for inflation), followed by biotechnology firms (41%) and pharmaceutical companies (15%).

When compared with the previous study, biomedical research expenditures by industry have decreased from a compound annual growth rate (adjusted for inflation) of 8.1% from 1994-2003 compared with 5.8% for 2003-2007 (P = .05). From 2003-2008, the number of new and novel drug and device approvals did not increase (Table 2).

Total biomedical research expenditures at universities were $27.7 billion in 2007, up from $22.0 billion in 2003. When adjusted for inflation, this represents an increase of 7.8% from 2003 to 2007. Federal sources remain the largest contributor to academic biomedical research expenditures, accounting for 65% of expenditures, followed by institutional funds (18% of expenditures).

The distribution of NIH funding among the most heavily funded institutions has remained almost unchanged from 1994 to 2007. In 2007, the 10 most heavily funded institutions received 20% of NIH funding (compared with 19% in 1994), and the top 50 institutions received 58% of NIH funding (compared with 55% in 1994).

The federal government and foundations spent $1.8 billion in 2003 and $2.2 billion in 2008 (nominal values) on health services research. The primary sources of federal funding for health services research were the NIH ($1.0 billion in 2008) and the Agency for Healthcare Research and Quality ($335 million in 2008). The Robert Wood Johnson Foundation contributed $317 million to health services research in 2003 and $523 million in 2008.

Since 2005, the rate of increase of research spending has slowed, with a compounded annualized growth rate of 3.4% (2003-2007) vs 7.8% (1994-2003). Total spending on biomedical research accounted for approximately 4.5% of total US health expenditures in 2007. While the decrease has occurred at a time of intense economic instability and financial upheaval in the world's financial markets, the rate diminished even before the events of 2007-2008. Funding from the NIH and industry, which includes pharmaceutical, biotechnology, and medical device firms, slowed from 2003 to 2007 and, after adjusting for inflation, has decreased in 2008. Therefore, research investment appears to have returned to its previous cyclical pattern of increases noted since the 1940s.²¹ The rate and cyclic nature of sponsorship affects researchers and institutions, because it influences career choice, selection of projects, building of laboratories, and establishment of new programs. It makes them cautious and may portend a trend to favor incremental research rather than high-risk/high-reward avenues, which have particular value to refractory diseases and those of great clinical or public health impact.

Companies continue to supply the largest proportion of total research spending (58%). However, the rates of growth vary by type of company, with medical device firms eclipsing biotechnology companies and both surpassing conventional pharmaceutical firms. Over this same period, the return as measured by stock price appreciation was greatest for the medical device sector (Figure 2). In turn, stock performance influences companies' overall ability to invest as well as the choice between low-risk, near-term projects vs high-risk, long-term undertakings. These findings are consistent with our 2005 analysis, although biotechnology companies replaced device companies as leaders of total shareholder return. These companies' stock performance vis-à-vis conventional pharmaceuticals likely reflects several factors that influence the investor's perception of value: an easier route to FDA approval; lower cost of trials, favorable pricing and eventual profitability of the commercial product; and predictable demand and direct marketing channels through a limited number of practitioners. Thus, devices and bioengineered drugs are both judged to be better company investments, a trend since 1994 that we can confirm.

Furthermore, a higher proportion of companies' spending is for late-stage clinical trials rather than drug discovery.¹ Their preference for investments of lower risk is also reflected in accelerated purchasing of small biotechnology firms by large pharmaceutical companies. Because of these trends, some observers have predicted that the industry model is changing, with large companies buying small ones rather than investing in (and conducting) early stage, discovery research themselves.²² This is clearly problematic in the absence of increased research investment pools for the small company, because this is the most common route for academic research to enter clinical use.²³ It highlights the important role of foundations and other private research sponsors, because they fill the gap between government-sponsored and commercial research.

Funding by foundations and charities also slowed from 2003-2007 compared with a decade earlier. These institutions were especially affected by the 2007-2009 recession, a time when their ability to fund speculative, high-risk research is particularly needed. With the intent of enhancing the effectiveness and productivity of grants, foundations are exploring alternative research models, such as joint funding with industry and the NIH, freestanding institutes outside academia, offshore (low-overhead) contracting, and pay-for-performance contracts in lieu of conventional grants.²⁴

This study presents a US perspective. However, previous analyses indicate that about 70% to 80% of total global biomedical research and development is sponsored by the US public sector, US-based foundations, and US-headquartered corporations.²⁵ This is in contrast to other research and development, for which the US sponsorship accounts for only about one-third.²⁶ As a proportion of total health care spending, the United States invests 4.5%, an amount higher than for any other country.²⁵ An exception is the US support for health services research, which accounts for only 2% of total research or 0.1% of total US health care spending and which receives more funding in Europe.²⁷

The estimates for 2003 and 2004 have been revised downward from our previous study, owing to improved data and analyses that reduce double counting of research and development support from industry. However, although the absolute amounts have been revised, limiting the ability to compare the magnitude of the totals from 1994-2003 to 2003-2007, the comparison of relative trends remains valid. The selection of 2003 as the point for comparing the change in the trends in funding is arbitrary and reflects the time at which the analyses were conducted. Because of differences in the data sources between the periods, the exact year in which the rate of funding changed cannot be identified. However, the objective of this study was to quantify the funding for biomedical research from 2003-2007 and to compare changes in this period with those observed from 1994-2003. In this analysis, data are improved but still have limitations. To quantify total funding across all major sponsors, we had to rely on disparate sources that may not be directly comparable. In addition, estimates of biomedical research funding are likely conservative, because the data do not capture all sources (eg, private philanthropy) and are not exhaustive (eg, exclude small device firms).

This analysis suggests that market value of different industry sectors move in parallel with research investment, which has driven the strikingly favorable performance of the medical device sector from 1994 to 2003 and the medical biotechnology (chiefly large-molecule drugs) sector since 2003. Performance of conventional pharmaceuticals (predominantly small-molecule drugs) has decreased in comparison. Many reasons can explain this performance lag, including higher regulatory hurdles, longer and more expensive clinical trials, higher failure rates in preclinical studies, and less flexibility in pricing. When conventional small-molecule drugs are compared with large-molecule agents that have known biological actions or with devices that rely on minor engineering changes, the incentives for drugs are less attractive. For this reason, some investigators have suggested new incentives for new drugs that are effective against diseases of high societal burden or gravity for the individual. These include incentives for novelty and comparative effectiveness, extended patent life, and pricing enhancements for drugs in particular need. Such incentives have been used successfully for vaccines and low-prevalence “orphan” diseases.²⁸

The number of new drug or device approvals is an incomplete measure of research productivity. Broader health status measures favorably reflect the result of biomedical research investments. These include lower death rates for cancer, stroke, and heart disease; longer life expectancy and improved quality of life for persons older than 65 years; more effective and earlier disease detection; substitution of noninvasive for invasive interventions; and improved pain avoidance and control. Notably, few of these rely on the model of one drug for one disease; many reflect public health advances, not solely new technology.²⁹

The cost of care in the current era will be an even greater influence on research investment than it has been in the era just concluded. The United States and other developed countries have aging populations, greater burdens of chronic diseases, an increased sense of obligation toward disease in the developing world, and new or refractory infections (eg, 2009 influenza A[H1N1]) of public health import. Economic limitations are also palpable as the United States considers alternatives to private insurance, limits to public funding, and more realistic actuarial assumptions. Moreover, all countries are facing nearly identical rates of total spending increase (albeit from very different base amounts as a percentage of GDP). Therefore, the cost and value of technology will likely receive additional scrutiny. Hence, it will be even more important to examine research productivity critically.

What might improve researchers' productivity and their effectiveness? Changes recently recommended include routine interchange of ideas and individuals between laboratories having complementary capabilities; less onerous patents for basic discoveries to promote access and research use; recruitment of scientists from other fields, especially from informatics and information sciences, materials, and physics; and the systematic search for unconventional approaches to refractory research problems. Such remedies have motivated the NIH Roadmap programs, reorganization of internal research and development efforts within industry, and experimentation by foundations with unconventional alliances and grants. Many of these changes are aimed at creating “semipermeable membranes” between laboratories within and outside different institutions, as well as between companies and universities.³⁰

New technology can be viewed either as an undesirable cost or as a source of value. In the countries within the Organisation for Economic Co-operation and Development (comprising the 8 largest economies), between 30% and 45% of the yearly increase in medical spending is attributed to new technology.²⁷ Some observers in the United States and Europe consider that rate of increase to be unsustainable (as the population ages) and an undesirable drain on overall economic activity—each 1% increase in medical spending lowers GDP by 0.3%.³¹ Others counter that view and see investment in health care and technology as sources of a country's competitive advantage rather than a financial drain.³² Therefore, it is inevitable that incentives for development of new medical technology will be scrutinized. Also, because the amount spent on health services research and effectiveness and on clinical epidemiology is so much less than that spent on new technology, pressure will likely mount to direct funds toward new tools to evaluate the clinical value of technology.³³

Biomedical research captures the public's imagination. It serves many masters and is highly valued as a source of new and more effective treatments for common or devastating diseases. Research, and the products and services it leads to, are sources of economic development, which is recognized in the developed and developing world alike. Therefore, in the coming years, debate will likely increase between those who view technology as a source of additional cost and those who view it as a source of value. The research community should be mindful of how others view it and take aggressive steps to enhance its own productivity.

Corresponding Author: E. Ray Dorsey, MD, MBA, 1351 Mt Hope Ave, Ste 223, Rochester, NY 14620 (ray.dorsey@ctcc.rochester.edu).

Author Contributions: Dr Dorsey had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Dorsey, Moses.

Acquisition of data: de Roulet, Thompson, Reminick, Thai, White-Stellato, George.

Analysis and interpretation of data: Dorsey, de Roulet, Thompson, Reminick, Thai, White-Stellato, Beck, George, Moses.

Drafting of the manuscript: Dorsey, de Roulet, George, Moses.

Critical revision of the manuscript for important intellectual content: Dorsey, de Roulet, Thompson, Reminick, Thai, White-Stellato, Beck, George, Moses.

Statistical analysis: Thompson, Beck.

Obtained funding: Dorsey.

Administrative, technical, or material support: Dorsey, Moses.

Study supervision: Dorsey, Moses.

Funding/Support: This study was funded by grant KL2 RR024136 from the National Center for Research Resources, a component of the National Institutes of Health (NIH), and the NIH Roadmap for Medical Research.

Role of the Sponsor: The funding sources had no role in the design and conduct of the study; the analysis or interpretation of the data; or the preparation or final approval of the manuscript.

Additional Contributions: We thank Aaron Catlin, MS, of the Centers for Medicare & Medicaid Services, for his assistance in understanding changes in the National Health Expenditure data. Mr Catlin received no extra compensation for his assistance.

This article was corrected online for typographical errors on 1/12/2010.