With the founding of the National Institute of Biomedical Imaging and Bioengineering (NIBIB) in 1999, the National Institutes of Health (NIH) made explicit its dedication to expanding research in biomedical engineering. Ten years later, we sought to examine how closely federal funding for biomedical engineering aligns with U.S. health priorities. Using a publicly accessible database of research projects funded by the NIH in 2008, we identified 641 grants focused on biomedical engineering, 48% of which targeted specific diseases. Overall, we found that these disease-specific NIH-funded biomedical engineering research projects align with national health priorities, as quantified by three commonly utilized measures of disease burden: cause of death, disability-adjusted survival losses, and expenditures. However, we also found some illnesses (e.g., cancer and heart disease) for which the number of research projects funded deviated from our expectations, given their disease burden. Our findings suggest several possibilities for future studies that would serve to further inform the allocation of limited research dollars within the field of biomedical engineering.
This is a preview of subscription content, log in via an institution to check access.
Access this article Subscribe and saveSpringer+ Basic
€34.99 /Month
Price includes VAT (Germany)
Instant access to the full article PDF.
Similar content being viewed by others Explore related subjectsDiscover the latest articles and news from researchers in related subjects, suggested using machine learning. ReferencesAmerican Association for the Advancement of Science. Table. NIH Budgets by Institute and Funding Mechanism. AAAS, 2008. Available at: http://www.aaas.org/spp/rd/health09p.pdf. Accessed 8 March 2010.
Biomedical Engineering Source. Available at: www.bmesource.org. Accessed 8 March 2010
Black, D. A., and J. D. Pole. Priorities in biomedical research: indices of burden. Brit. J. Prevent. Soc. Med. 29:222–227, 1975.
Bronzino, J. D. Introduction and preface. In: The Biomedical Engineering Handbook, 2nd edn, edited by J. Bronzino. Boca Raton: CRC Press, 2000.
Cohen, J. W., and N. A. Krauss. Spending and service use among people with fifteen most costly medical conditions. Health Aff. 22(2):129–138, 1997.
Committee on NIH Research Priority-Setting Process. Scientific Opportunities and Public Needs: Improving Priority Setting and Public Input at the National Institutes of Health. Washington, DC: National Academy Press, 1998.
FY 2005 Budget in Brief: National Institutes of Health. Washington, DC: U.S. Department of Health and Human Services, 2004.
Griffith, L. G., and A. J. Grodzinsky. Advances in biomedical engineering. J. Am. Med. Assoc. 285(5):556–561, 2001.
Gross, C. P., G. F. Anderson, and N. R. Powe. The relation between funding by the National Institutes of Health and the burden of disease. N. Engl. J. Med. 340(24):1881–1887, 1999.
Hendee, W. R., S. Chien, C. D. Maynard, et al. The National Institute of Biomedical Imaging and Bioengineering: history, status, and potential impact. Ann. Biomed. Eng. 30:2–10, 2002.
Kirschstein, R. Disease-specific Estimates of Direct and Indirect Costs of Illness and NIH Support: Fiscal Year 2000 Update. National Institutes of Health, Office of the Director, 2000.
Moses, III, H., E. R. Dorsey, D. H. Matheson, and S. O. Thier. Financial anatomy of biomedical research. J. Am. Med. Assoc. 294(11):1333–1342, 2005.
The National Institutes of Health. Institutes, Centers, and Offices. NIH, 2008. Available at: http://www.nih.gov/icd/index.html. Accessed 8 March 2010.
The NIH Task Force of the Bioengineering Division, Basic Engineering Group, Council on Engineering. Position Statement on the FY 2006 Budget Request for the National Institutes of Health. American Society of Mechanical Engineers, Council on Engineering, 2005.
Neumann, P. J., A. B. Rosen, D. Greenberg, et al. Can we better prioritize resources for cost-utility research? Med. Decis. Mak. 25(4):429–436, 2005.
Steinbrook, R. Health care and the American Recovery and Reinvestment Act. N. Engl. J. Med. 360:1057–1060, 2009.
Success Rates for all NIH Competing Grant Applications by NIH Institutes/Centers, Grant Mechanism, and Activity Codes. NIH Office of Extramural Research: Fiscal Year 2008.
Working Group on Priority Setting. Setting research priorities at the National Institutes of Health. Bethesda, MD: National Institutes of Health, 1997.
World Health Organization. The World Health Report 2002: Reducing Risks, Promoting Healthy Life. Geneva: World Health Organization, 2002.
School of Public Health, Yale University, New Haven, CT, USA
Jessica B. Rubin & A. David Paltiel
Department of Biomedical Engineering, Yale University, MEC413, 55 Prospect Street, New Haven, CT, 06511, USA
Jessica B. Rubin & W. Mark Saltzman
Correspondence to W. Mark Saltzman.
Additional informationAssociate Editor Scott I. Simon oversaw the review of this article.
Appendices Appendices Appendix A NIH Institute abbreviations13 Technical Appendix Measuring Disease BurdenThe three measures of U.S. disease burden used in this analysis were: (1) cause of death; (2) disability-adjusted survival loss; and (3) health expenditure. These measures were obtained from two sources. The World Health Organization’s World Health Report 2002 provided data on measures (1) and (2) using approximately 150 specific diseases. To improve manageability, these data were rolled into 65 major categories provided by the Report. A sample of this data is seen in Table A1.
Table A1 Example of 2002 World Health Organization U.S. disease burden dataA 1997 analysis by Cohen and Krauss provided cost information for the 15 most expensive diseases in the U.S. using different disease categories than the World Health Report. These data (see Table A2) were used as measure (3) of disease burden. Though the top 15 conditions account for over 50% of all U.S. health care expenditure, the rest of the expenditure categories would have provided interesting data as well.
Table A2 Cohen and Krauss5 U.S. health expenditure data Identifying Grants Using CRISP DatabaseThe Computer Retrieval of Information on Scientific Projects (CRISP) database was used to identify federally funded biomedical engineering grants. Searches were restricted to “research projects,” which included grants with activity codes R01, P01, U01, R23, R29, R35, R37, R03, R15, R12, U19, or R55. In order to limit the sample to a manageable size, only research projects from the year 2008 were considered.
We first searched 2008 research projects from all NIH Institutes that matched the exact search term biomedical engineering. This only included projects funded by the 20 NIH-defined Institutes and not the seven NIH Centers (e.g., Center for Information Technology, Center for Scientific Review), as an initial search including projects funded by these centers provided only an additional three projects, which were not traditional laboratory-based research. We chose the more restrictive search term biomedical engineering after the broader term bioengineering produced over 1000 results, many of which were not applicable to the field. Additional relevant search terms, such as biomedical imaging, biomaterial, device, biomedical instrumentation, and tissue engineering were considered but rejected when they showed signs of expanding the final dataset in a direction that was skewed toward particular types of research. Though projects that matched the search term biomedical engineering did not produce an all-inclusive set, they provided the most representative sample, a set of 242 projects in all research areas and disease categories distributed among the NIH Institutes.
A subsequent search identified all 2008 research projects funded by the National Institute of Biomedical Imaging and Bioengineering, producing a second sample set of 467 projects. Thirty-eight of these projects also matched the search term biomedical engineering, and thus were included in both the first and second dataset. A total of 671 unique biomedical engineering projects were identified using these two search methods. After excluding 30 additional projects with either the same title but different grant numbers (primarily supplements), or projects for which an abstract was not provided, a resulting 641 projects were analyzed.
Classifying Grants into Research CategoriesOur original intent had been to classify all research projects into disease categories in order to compare research to disease burden. However, many of the projects identified in the two datasets (52%) were not disease-specific. Rather, they were general technique development studies applicable to multiple disease areas. In order to obtain a more complete and accurate picture of federally funded biomedical engineering research both within the NIBIB and throughout the NIH, all of the projects in both datasets were classified into one of 16 specific research categories within the field of biomedical engineering, such as imaging, tissue engineering, and drug delivery. The research categories were defined using three independent sources: a biomedical engineering textbook,8 a published review of the field,4 and a biomedical engineering informational source2 (see Table A3). One contained categories that were not mutually exclusive, one contained categories that were too broad, and in one the categories were not all-encompassing. All three sources provided detailed information about inclusions in each of our categories (see Table A4).
Table A3 Biomedical engineering research categories as defined by three independent sources Table A4 Biomedical engineering research category classifications Classifying Grants into Disease CategoriesFor the 48% of projects that were disease-specific, each was classified into both the World Health Organization and the Cohen and Krauss disease categories. Projects were considered disease-specific if the abstract described how the investigators applied their technology to a particular disease. If the disease was merely mentioned in the abstract of the study as a potential application of the research, it was not included. Information about the projects could only be obtained using the abstracts provided by the researchers, a limitation of using the CRISP database.
A project could be classified into more than one disease category only if, in the abstract, the researcher described how the study would impact two distinct diseases. We justify this classification approach by noting that our study was a relative analysis, not an absolute one. We sought to determine whether the number of research projects being funded in each disease area was more or less than expected based on disease burden. Thus, we sought to identify trends, rather than to quantify results. If the abstract describing a project made specific mention of applying a given technology in the treatment of both brain tumors and stroke, that project was counted in both categories. Double counting (which only occurred for 6% of projects) did not change the relative standing of each disease category. Moreover, because it did not require us to make additional assumptions about either the value of each disease category or how project resources were allocated across disease categories, this method of counting might be judged more representative than a method that arbitrarily assigned half value to each disease category. It is important to note that if an abstract described a technique, and noted that in the future it might be applicable to multiple different diseases, it was not counted for any of those disease categories. Additionally, when counting the total number of projects that were disease-specific, each project was only counted once (i.e., even if the project looked at both heart disease and cancer, it was only counted once when tabulating which projects were disease-specific). Thus, this issue does not affect the 48% and 52% figures about projects that were or were not directed toward particular diseases.
Some of the diseases with the highest burden using the World Health Organization categorizations were specific types of cancer, such as lung cancer and breast cancer. However, while many of the cancer studies conducted were specific to one of these types, some were general cancer research studies, directed at improvements in diagnosis or treatment for all types of tumors. Therefore, all cancer research projects that were type-specific were categorized into both their specific disease category and into the general cancer category.
Once all of the projects were classified into one or more disease categories for each of the disease classification methods, the total number of projects in each disease area was tabulated (see Table A5). These tallies were used in the data analysis (see next section).
Table A5 Number of 2008 NIBIB research projects classified by Cohen and Krauss disease categories Analyzing DataIn addition to determining the distribution of projects in each of the 16 research categories and funded by each of the 20 NIH Institutes, the disease category distributions of the projects were analyzed as well. The correlation between the number of projects in each disease category and levels of disease burden (using all three measures) was determined using Spearman’s rank correlation test. This statistical analysis allowed us to test for a correlation between these two variables without limitations regarding the function linking the variables. This procedure had been used in previous studies by Gross et al. and Neumann et al. Simple linear regression was then performed for the 15 most burdensome diseases using each of the three measures of disease burden. Regression-derived estimates were calculated to predict the expected number of grants in each disease area. For a selected five of the most burdensome diseases, residuals were calculated from these estimates (observed number of grants minus expected number of grants) in an effort to identify outliers among all three measures of disease burden. Because the World Health Organization disease categories differed slightly from the Cohen and Krauss categories, a common set of five disease categories was established. For World Health Organization burden measures, heart disease was calculated as the sum of ischemic and nonischemic heart disease and mental disorders was calculated as the sum of all neuropsychiatric conditions.
About this article Cite this articleRubin, J.B., Paltiel, A.D. & Saltzman, W.M. Are We Studying What Matters? Health Priorities and NIH-Funded Biomedical Engineering Research. Ann Biomed Eng 38, 2237–2251 (2010). https://doi.org/10.1007/s10439-010-9982-9
Received: 08 January 2010
Accepted: 20 February 2010
Published: 12 March 2010
Issue Date: July 2010
DOI: https://doi.org/10.1007/s10439-010-9982-9
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4