Georgetown University Law Center Scholarship @ GEORGETOWN LAW

2012

The Limits of Government Regulation of Science

John D. Kraemer Georgetown University Law Center, [email protected]

Lawrence O. Gostin Georgetown University Law Center, [email protected]

Georgetown Public Law and Legal Theory Research Paper No. 12-020

This paper can be downloaded free of charge from: https://scholarship.law.georgetown.edu/facpub/776 http://ssrn.com/abstract=2012429

Science, March 2012, at 1047-1049

This open-access article is brought to you by the Georgetown Law Library. Posted with permission of the author. Follow this and additional works at: https://scholarship.law.georgetown.edu/facpub Part of the Constitutional Law Commons, First Amendment Commons, Health Law and Policy Commons, and the Science and Technology Law Commons

P UBLIC H EA LTH A ND BIOS EC URIT Y

A transparent institutional review process The Limits of Government will balance scientific freedom and national security better than publication restrictions. Regulation of Science

John D. Kraemer1,2 and Lawrence O. Gostin2*

ast summer, two research teams ology and results could become a blueprint about building a hydrogen bomb, even though funded by the National Institutes of for bioterrorism (1). the information was in the public domain; the LHealth genetically modif ied H5N1 The U.S. government’s request not to injunction was later vacated when the article avian influenza vir uses, making them publish key scientific findings sparked con- was published elsewhere (2). In 2005, the capable of eff icient respiratory transmis- siderable controversy. To many research- Proceedings of the National Academy of Sci- sion between ferrets. Ferrets are thought ers, knowledge about what mutations enable ences refused to comply with an HHS request to be a good animal model for influenza in respiratory transmission is essential to sur- to decline publishing a mathematical model humans. A small number of genetic changes veillance of and early action against vari- of botulism in the milk supply (3). The H5N1 might be able to convert the presently zoo- ants of H5N1. They worry that government case, however, is the first time government has notic H5N1 virus into a pathogen with dan- intrusion into scientif ic innovation would sought to redact information after an institu- gerous pandemic potential—transmissible discourage vital research. However, security tionalized HHS review process.

Constitutional Limits on Government The court ruled that federally funded scientific research, Restrictions of Scientific Publications especially at universities, should be free from prior restraint— The First Amendment to the U.S. Consti- tution affords considerable protection to calling into question the validity of CUI conditions on political, artistic, and scientific expression, that could trigger “strict scrutiny” by the research grants. Supreme Court (4). The court is most vig- orous in reviewing government restraints from human-to-human, with a >50% case- advocates believe the greater risk is that the on speech in advance of publication, which fatality rate. The National Science Advisory mutated virus could escape or that knowl- it calls “prior restraints.” Prior restraints are Board for Biosecurity (NSABB), which edge about these mutations could get into the uniquely threatening to First Amendment advises the U.S. Department of Health and wrong hands. They suggest that research of values because they prevent ideas from ever Human Services (HHS), recommended that this kind should not be funded or undertaken being heard (5). two journals, Science and Nature, redact in the first place. Where, as here, the research Had the government compelled the H5N1 key information before publication. The has already been conducted, they urge sci- researchers to cease research or the journals NSABB and HHS expressed concerns that entific journals not to publish any sensitive to withhold publication—whether through published details about the papers’ method- methods or results (1). the force of law or by creating adverse con- The HHS request reveals a troubled rela- sequences such as loss of funding—it could 1Department of Health Systems Administration, George- tionship between security and science. This is have violated the First Amendment. Even town University School of Nursing and Health Studies, not the first time a government has requested informal systems of restraint can be uncon- Washington, DC 20057, USA. 2O’Neill Institute for National and Global Health Law, Georgetown University Law Center, that a journal not publish information. In stitutional, such as a government threat to Washington, DC 20001, USA. 1979, the U.S. Department of Energy secured prosecute publishers (5). In this case, how- *Author for correspondence. E-mail: [email protected]. an injunction against the magazine The Pro- ever, HHS’ request, by its own terms, was edu gressive to prevent the publication of an article nonbinding, and the journals had discre-

www.sciencemag.org SCIENCE VOL 335 2 MARCH 2012 1047 Published by AAAS

POLICYFORUM

tion whether or not to comply (6). Given on research grants. The wider the scope of lication edits without a specific license. In the absence of legal force or undue induce- CUI conditions, the more likely that courts essence, OFAC argued that editing a paper ments or penalties, the government’s request will invalidate them (4). was providing a service to foreign authors in to withhold information does not violate the The Supreme Court’s “unconstitutional violation of trade embargoes. In 2004, OFAC First Amendment. conditions” doctrine holds that government reversed that decision and allowed normal There are situations in which a govern- may not place conditions on public fund- scientific editing to occur (14). Had OFAC ment has the authority to block scientif ic ing that require the recipient to surrender not reversed itself, First Amendment chal- communications. The clearest case is when First Amendment rights. Thus, government lenges against the policy likely would have research has been properly classified under has no obligation to provide research fund- prevailed (15). federal law and the person seeking to com- ing, but if it chooses to, it cannot restrain the municate f indings obtained it under the free expression of researchers without a com- Access to Sensitive Data Under the Freedom terms of a security clearance—whether they pelling state interest. For example, a federal of Information Act are still working for the government or not, appellate court recently struck down HHS A functioning democracy requires that citi- so long as procedural requirements are met guidelines requiring recipients of AIDS pre- zens be able to access information in the (7). Although a researcher is obliged to keep vention funding to pledge their opposition to government’s possession, but not if access classified information confidential, publish- prostitution, reasoning that it was an uncon- poses an unacceptable security risk. The ers who obtain that information lawfully have stitutional condition (10). Freedom of Information Act (FOIA) bal- a right to publish. In the Pentagon Papers The unconstitutional conditions doctrine, ances these concerns by affording access to case, the Supreme Court held that President however, is hard to decipher. For example, the federal agency records unless the records Nixon did not overcome the “heavy presump- Supreme Court upheld HHS prohibitions on fall within a statutory exemption. Federal agencies support much of the research in the United States, including both of the recent Could the public obtain sensitive data that have been redacted H5N1 studies. Could the public obtain sensi- tive data that have been redacted from publi- from publications through a FOIA request? If so, governmental cations through a FOIA request? If so, gov- requests to redact sensitive information would be fruitless. ernmental requests to redact sensitive infor- mation would be fruitless. FOIA applies only to “agency records,” tion” against prior restraint when he sought the use of family planning funds to counsel so a threshold issue is whether university to prohibit publication of classified materials. women regarding abortion, reasoning that research data acquired under a grant constitute The court found that an undefined concept of government is entitled to subsidize one pro- an agency record. In 1980, the Supreme Court “security” did not “abrogate the fundamental tected right (family planning), while refusing ruled that research data produced under an law embodied in the First Amendment” (8). to subsidize analogous rights (abortion coun- NIH grant and used in regulatory proceedings It is far less clear whether government seling) (11). The court similarly upheld the by the U.S. Food and Drug Administration may suppress the publication of research con- government’s right to withhold funding to any did not constitute an agency record subject ducted with government funding when the public university that denied access to mili- to FOIA because it was retained by the non- results are “controlled unclassified informa- tary recruiters, even though the universities governmental grantee. The court found that tion” (CUI) [sometimes referred to as “sen- claimed it violated their freedom to disap- FOIA required the agency to either produce sitive but unclassified” (SBU)] under con- prove of the military’s “don’t ask, don’t tell” or obtain permanent custody of the data (16). ditions set by government grants or con- rule. The court said the law neither denied the The “Shelby amendment,” enacted in tracts. Traditionally, the federal government institutions the right to speak nor required 1999, expanded public access to data pro- restricted communication about basic sci- them to say anything (12). duced at universities and other nonprof it ence research only through classif ication. research entities under federal grants. The However, CUI restrictions have become more Scientific Publication from Countries public can request the data if they were pro- common, and no court has directly addressed Subjected to U.S. Economic Sanctions duced under a federal grant and “cited pub- their constitutionality. Although it is unclear In the past, the federal government has lically and officially by the Federal Govern- how often CUI clauses include a prepublica- impeded scientif ic publication processes, ment in support of an agency action that has tion review requirement, research suggests not because of articles’ content but rather the force and effect of law” (17). Federal agen- that they occur with some regularity (9). because the authors were from countries cies could take care not to officially cite highly Board of Trustees of Leland Stanford Jr. against which the United States had imposed sensitive data, thereby avoiding a successful University v. Sullivan is the most pertinent economic sanctions. The Department of FOIA request. However, it is not always sim- case for evaluating CUI restrictions. Stanford Treasury’s Office of Foreign Assets Control ple or easy to refrain from referencing sensi- University challenged an NIH confidential- (OFAC) enforces these economic sanctions. tive research. The NIH, for example, might ity clause that required the university to seek For a brief period in 2003, OFAC restricted the reasonably refer to the H5N1 research as jus- prior approval before publishing preliminary review process for scientific papers submit- tification for revising biosecurity policies. findings about artificial heart research to pro- ted from countries sanctioned by the United Even if sensitive data do become part of tect the public from unvalidated research find- States (13). In particular, OFAC informed the an agency record, FOIA provides the federal ings. The court ruled that federally funded Institute of Electrical and Electronics Engi- government with ample authority to refuse scientific research, especially at universities, neers (IEEE) that, although its journals could a request on security grounds. FOIA pro- should be free from prior restraint—calling subject papers from sanctioned countries to vides nine exemptions under which records into question the validity of CUI conditions peer review, they could not make prepub- that would otherwise have to be disclosed

1048 2 MARCH 2012 VOL 335 SCIENCE www.sciencemag.org Published by AAAS

POLICYFORUM

may be withheld, one of which is for “mat- The problem of government discretion is tional review. If IBCs are formally designated ters that are specifically authorized under cri- compounded by highly inconsistent practices to conduct the institutional review function, teria established by an executive order to be among federal agencies in the classification HHS will have to clarify whether NSABB kept secret in the interest of national defense systems they use. There is inconsistency of will guide and oversee the process (21). In or foreign policy and are in fact properly clas- structure (the labels attached, such as classi- addition, because IBCs may recommend sified pursuant to such an executive order” fied, CUI, SBU, or other terminology), as well that researchers voluntarily restrict access to (18). Through this exception, Congress has as in the application of that structure to indi- methods or results in some instances, it will acknowledged broad executive authority to vidual documents (no clear standard exists be important for HHS to develop a system for classify records so long as it is done lawfully for deciding whether to classify particular managing access to sensitive data and for dis- pursuant to an executive order. information). In short, the line between clas- seminating it to those with a need to know in President Obama’s 2009 Executive Order sified and CUI remains unclear, as agencies a fair manner. 13526 revises existing classification stan- struggle to apply President Obama’s execu- If HHS improves its functioning, the insti- dards (19). Although it was designed to tive orders (24). tutional review process can ensure a sound reduce the amount of classified materials, balance between scientif ic freedom and the executive order affords agencies con- Balancing Scientific Freedom, Constitutional national security. A fair, transparent process siderable discretion to classify on security Values, and Biosecurity undertaken by research institutions, with a grounds. Consistent with prior policy, the The federal government has the power to balanced approach to scientific benefits and executive order mandates that “basic scien- prevent the dissemination of sensitive life- public safety, together with HHS guidance tific research information not clearly related sciences research, but there are good rea- and oversight of high-risk research, is prefer- to the national interest shall not be classi- sons to exercise that power sparingly. The able to government constraints on scientific fied.” However, the order permits the classi- current system of deliberation by a federal information by force of law. fication of “scientific, technical, or economic expert advisory board and HHS-issued vol- matters relating to the national security,” pro- untary recommendations is preferable to for- References and Notes 1. D. Grady, W. J. Broad, New York Times, 21 December 2011, vided that disclosure is reasonably expected mal government mandates. Although we do p. A1. “to cause identifiable or describable damage not have all the data, the NSABB process in 2. United States v. Progressive, Inc., 467 F. Supp. 990 (1979). to the national security.” Furthermore, agen- the H5N1 cases appears reasonable, given 3. Letter from Stewart Simonson to Bruce Alberts, 27 May cies may classify data that meet the execu- that unredacted publication could enable bad 2005; www.fas.org/sgp/bush/hhs052705.pdf. 4. Board of Trustees of Leland Stanford Jr. University v. tive order’s standards even if the data were actors with scientif ic skill to replicate the Sullivan, 773 F. Supp. 472 (1991). not classified at the time of the FOIA request studies, with profoundly harmful effects. The 5. Bantam Books v. Sullivan, 372 U.S. 58 (1963). (19). Thus, federal agencies have wide federal government has promised to share the 6. NIH, Press statement on NSABB review of H5N1 Research (20 December 2011); www.nih.gov/news/health/dec2011/ authority to prevent the release of research researchers’ methods and conclusions with od-20.htm. information through a FOIA request simply scientists with a need to know, which substan- 7. Snepp v. United States, 444 U.S. 507 (1980). by classifying it, provided that there are legit- tially advances scientific objectives. 8. N.Y. Times v. United States, 403 U.S. 713 (1971). 9. A. Chamberlain, Science and Security in the Post-9/11 imate national security justifications. Can the review process for high-risk bio- Environment—Export Controls: Grants, Contracts, and In 2010, President Obama issued a fur- logic research be improved further? The Publishing (American Association for the Advancement ther executive order stating that CUI is not NSABB’s origins can be traced to the so- of Science, Washington, DC, 2004); www.aaas.org/spp/ post911/grants. automatically exempt from FOIA (20). Thus, called Fink report issued in 2004 by the 10. Alliance for Open Society International v. USAID (2nd Cir. to ensure that sensitive biological research National Research Council (21). However, 2011). information is not disclosed, agencies would vital aspects of the Fink report have not been 11. Rust v. Sullivan, 500 U.S. 173 (1991). 12. Rumsfeld v. Forum for Academic and Institutional Rights, have to classify it. [Certain nonbiological implemented. In particular, the Fink report Inc., 547 U.S. 47 (2006). research, such as nuclear energy, is automati- proposed an institutional review process for 13. Letter from R. Richard Newcomb (OCAF) to Nelson cally exempt from FOIA, as are the locations biological “experiments of concern”—those G. Dong, 2 April 2004; www.ieee.org/documents/ OFAC_40204_Letter.pdf. where select biological agents are held (21).] falling into seven research classes, making 14. IEEE, OFAC Background; www.ieee.org/about/corporate/ Some research data also might be protected the pathogen considerably more attractive compliance/regulatory/ofac/background.html. under FOIA exemptions for trade secrets as a bioterrorism agent (e.g., by enhancing 15. T. J. Chin, New York Univ. Law Rev. 83, 1883 (2008). 16. Forsham v. Harris, 445 U.S. 169 (1980). or predecisional deliberative memoranda virulence or transmissibility or by rendering 17. Office of Management and Budget, Revised circular 110A; within the government, but these options are vaccines ineffective). This approach was pat- www.whitehouse.gov/omb/circulars_a110. limited (22). terned on the Institutional Biosafety Commit- 18. Freedom of Information Act, 5 U.S.C. §552. 19. Executive Order 13526, Classified National Security Infor- The law, then, draws a distinction between tees (IBCs) required by NIH for recombinant mation (2009). classified and controlled unclassified infor- DNA research at institutions receiving fed- 20. Executive Order 13556, Controlled Unclassified Informa- mation. However, from a constitutional per- eral funding, which generally have been con- tion (2010). 21. National Research Council, Biotechnology Research in an spective, it would be troubling if the result sidered to be successful (21). Age of Terrorism (National Academies Press, Washington, turned solely on the label the government HHS, in partnership with institutions, DC, 2004). placed on the data. If the result did turn on the will have to ensure that the IBC model works 22. D. G. Goldenson, Boston Coll. Environ. Aff. Law Rev. 29, 311 (2001–2002). label, the government could simply relabel effectively: (i) institutions must develop the 23. M. Steffan, The high cost of keeping a secret, Washington research from CUI to classified and thus pro- requisite expertise to review dual-use research Post, ideas@innovation [blog], 13 October 2011. hibit its dissemination. Although decisions to of concern; (ii) HHS must specify the cat- 24. J. P. Fitzpatrick, M. A. Pustay, Guidance regarding classify can be challenged, prevailing is dif- egories of research requiring institutional Controlled Unclassified Information and the Freedom of Information Act [memorandum], 22 November 2011; ficult, and unnecessary classification is com- review—minimally including the seven types www.fas.org/sgp/cui/foia112211.pdf. mon (23). This appears to place too much dis- of high-risk experiments; and (iii) HHS must cretion in the hands of public officials. set clear and consistent standards for institu- 10.1126/science.1219215

www.sciencemag.org SCIENCE VOL 335 2 MARCH 2012 1049 Published by AAAS