2073 T here is substantial divergence between the scientific community’s standards for ensuring research integrity and the ad hoc protections for researcher independ- ence tolerated by federal regulatory agen- cies. The biomedical community’s concern about potential conflicts of interest is ad- dressed in the widespread (1, 2) policy of journals to require that authors of submitted articles disclose financial relationships so that editors and readers can judge whether conclusions might have been influenced by those financial ties. The editors of 13 leading biomedical journals have gone further and declared that they will no longer publish ar- ticles based on studies done under contracts in which the investigators did not have the unfettered right to publish the findings (1). With the increased involvement of uni- versities in commercial enterprises and collaborations, conflicts-of-interest con- cerns at academic institutions have grown in importance. In response, many institu- tions have implemented policies that at- tempt to ensure independence and protect the ability of researchers to share data with fellow scientists and the public (3–6). Research independence is also of great importance to regulators. Federal agencies charged with protecting the public’s health rely out of necessity on scientific evidence submitted by private parties in determining the hazardous characteristics of products and wastes. At the same time, there is growing evidence of conflicts of interest in private research submitted for regulation. For example, there are reports of a “fund- ing effect,” with sponsorship associated with favorable findings (3, 7, 8). There are also accounts of improper sponsor control over the design and reporting of results, and sponsor suppression or termination of research showing adverse effects (9–13). Except for limited prohibitions against the suppression of adverse effects, however, the quality and independence of private re- search used for regulation is subject to con- siderably less oversight than corresponding federally funded research. Most signifi- cantly, private research submitted for regu- latory purposes escapes external scrutiny if the research or the chemical under study is claimed to be confidential business infor- mation (14). Most of the applications sub- mitted to the U.S. Environmental Protection Agency (EPA) to market new chemicals, for example, contain science-relevant infor- mation that industry claims is confidential. Many of these trade secret claims do not appear to be justified (15). Yet without this information, it is not possible to evaluate the regulators’ decisions. Even when sponsored research is not pro- tected as trade secrets, the data underlying privately submitted research used for regula- tion need not be made publicly available, as is required for its federally funded counter- part (16). Also in contrast to public research, private research is not subject to the scientif- ic misconduct regulations promulgated by the U.S. Office of Research Integrity (17). Finally, even the “Data Quality Act”, which ostensibly is an attempt to improve the qual- ity of regulatory science through a formal complaint process, exempts a great deal of private research from its coverage (18). Despite the evident value of transparency about sponsorship in regulatory science, the disclosure of sponsor influence is generally not required or even requested by federal regulatory agencies. The EPA, the Occu- pational Safety and Health Administration, the Mine Safety and Health Administration, the Consumer Product Safety Commission, and the National Highway Traffic Safety Administration have no formal mechanisms to identify potential conflicts of interest, nor do they provide any incentive to encourage the conduct of research that is free of spon- sor control. The Food and Drug Admin- istration (FDA) has instituted a conflict pol- icy requiring financial disclosures for safety research conducted by private parties in sup- port of a license to market a drug or food ad- ditive (19). These disclosures do not, howev- er, distinguish between research where the sponsor controls the design or reporting of the research and research where sponsors have no control. Regulatory agencies should adopt, at a minimum, requirements for research inde- pendence comparable to those of biomedical journals. Disclosure of conflicts of interest should be required for all research, regard- less of whether it is federally or privately funded. Scientists should disclose whether they have a contractual right to publish their findings free of sponsor control and should identify the extent to which their work was reviewed by an affected party before publi- cation or submission to the agency. Sponsors who submit data should similarly disclose if their investigators had the contractual right to publish without sponsor consent or influ- ence. Finally, other parties (i.e., trade associ- ations, unions, or public interest groups) who submit scientific results should disclose all known conflicts of interests of the scien- tists conducting the studies. Regulators should not use conflict dis- closures to exclude research; they have the obligation to consider all evidence, accord- ing greater importance to studies of higher quality and relevance. Federal agencies should, however, develop policies that strongly encourage clear disclosures that counteract the strong incentives for spon- sors to influence research. Only then can agencies accurately weight studies and en- courage research independence. References and Notes 1. Uniform Requirements for Manuscripts Submitted to Biomedical Journals, www.ICMJE.org (October 2001). 2. F. Davidoff et al., JAMA 286, 1232 (2001). 3. S. Krimsky, Science in the Private Interest: Has the Lure of Profits Corrupted the Virtue of Biomedical Research? (Rowman & Littlefield, Lanham, MD, 2003). 4. U.S. General Accounting Office, “Biomedical research: HHS direction needed to address financial conflicts of interest” (GAO-02-89, Government Printing Office, Washington, DC, 2001). 5. M. K. Cho et al., JAMA 284, 2203 (2000). 6. American Assoc. of University Professors, Statement on Corporate Funding of Academic Research 2001, available at www.aaup.org/statements/ 7. J. E. Bekelman et al., JAMA 289, 454 (2003). 8. J. Lexchin et al., BMJ 326, 1167 (2003). 9. D. Rennie, JAMA 282, 1766 (1999). 10. S. A. Glantz et al., The Cigarette Papers (Univ. of California Press, Berkeley, 1996). 11. G. Markowitz, D. Rosner, Deceit and Denial: The Deadly Politics of Industrial Pollution (Univ. of California Press, Berkeley, 2002). 12. D. Rennie, JAMA 277, 1238 (1997). 13. P. Brodeur, Outrageous Misconduct: The Asbestos Industry on Trial (Pantheon Books, New York, 1985). 14. Code Fed. Regul. (C.F.R.) 40, §2.201–2.310. 15. U.S. General Accounting Office, Toxic Substances Control Act: Legislative Changes Could Make the Act More Effective (GAO/RCED-94-103, Government Printing Office, Washington, DC, 1994). 16. Pub. L. No. 105-277, Stat. 112, 2681 (1998). 17. C.F.R. 42, §50, A, Scientific Misconduct. 18. Fed. Regist. 67, 8452 (2002). 19. C.F.R. 21, §54. SCIENCE AND GOVERNMENT Disclosure in Regulatory Science David Michaels and Wendy Wagner D. Michaels is at the Department of Environmental and Occupational Health, George Washington University School of Public Health and Health Services, Washington, DC, 20052, USA. E-mail: eohdmm@gwumc.edu W. Wagner is at the University of Texas School of Law,Austin,TX 78705, USA. E-mail: wwagner@mail.law.utexas.edu P OLICY F ORUM www.sciencemag.org SCIENCE VOL 302 19 DECEMBER 2003