The full text of the letter is below and can be downloaded here.
July 11, 2018
Acting Administrator Andrew Wheeler
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, NW
Washington, D.C. 20460
Submitted electronically at www.regulations.gov
Re: Docket Number EPA-HQ-OA-2018-0259-0025, Strengthening Transparency in Regulatory Science
The Association of American Medical Colleges (AAMC), Association of American Universities (AAU), Association of Public and Land-grant Universities (APLU), and Council on Governmental Relations (COGR), collectively the “Associations,” write in response to the proposed rule issued by the Environmental Protection Agency (EPA), published in the Federal Register on April 30, 2018. The member institutions of AAMC, AAU, APLU, and COGR are the primary performers of federally funded research.
In this notice of proposed rulemaking (NPRM), the EPA has set forth a process that limits the science it will consider in critical rulemaking activities to those studies for which all underlying data are publicly available for analysis. Rather than foster the laudable goals of increasing transparency and enhancing the validity and reproducibility of scientific findings, this proposed rule would effectively prevent the EPA from evaluating the best available evidence when developing regulations specifically aimed at protecting human health. While our Associations strongly support transparency, reproducibility, and open science, we have never suggested that scientific research lacks merit or value if the data, for legitimate reasons, cannot be made publicly available or reproduced. We are particularly concerned when such rationale becomes the justification for ignoring scientific evidence that can save lives. The Associations, therefore, urge the EPA to withdraw this proposed rule.
The proposed rule does not advance the type of sound, evidence-based policymaking that is essential for every agency, and particularly important for the EPA, whose activities and regulations have a profound impact on air, land, and water quality, and thus the health of all Americans. This proposal thwarts the promise of evidence-based policymaking, squarely contradicting the requirement that the EPA use the “best available science” to make its regulatory decisions. Basing decision-making on only those studies with publicly available data would drastically curtail the use of key information and studies in the policymaking process and ignore the entire body of scientific evidence built up over years of inquiry.
The NPRM indicates that it “takes into consideration the policies and recommendations of third party organizations who advocated for open science.” However, while encouraging greater access to research data and suggesting incentives or necessary infrastructure enhancements to enable data sharing, most of the cited reports and policies from respected organizations listed do not recommend or support the premise that research should be disregarded if the data are not publicly available. The recommendations from one such report from the National Academies of Science, Engineering, and Medicine, “Optimizing the Nation’s Investment in Academic Research,” were incorporated into the 21st Century Cures Act and discuss the potential burdens of data sharing policies, but do not suggest that a policy such as the one being proposed by the EPA would be beneficial.
The NPRM further asserts that the policies are “informed by the policies recently adopted by some major scientific journals,” specifically citing the journals PLOS ONE, Science, and Nature. A commentary by the editors in chief of those journals in response to this NPRM refutes this characterization and rejects such a stringent approach, stating, “It does not strengthen policies based on scientific evidence to limit the scientific evidence that can inform them; rather, it is paramount that the full suite of relevant science vetted through peer review, which includes ever more rigorous features, inform the landscape of decision making. Excluding relevant studies simply because they do not meet rigid transparency standards will adversely affect decision-making processes.”
There are legitimate, reasonable and ethical reasons that scientific data may not be available to the public. This should not invalidate the research or its findings, and should not prevent the research from being used in important EPA rulemaking. In cases where it is not appropriate for data to be made publicly available, there are other mechanisms intrinsic to the scientific process for substantiating the relevance and validity of research results. Large-scale health and environmental studies generally involve sensitive data from human subjects, which may not be fully de-identifiable. In addition, many individuals agree to participate in these research studies through an informed consent process that ensures their data will not be shared in any form. The assurances provided to research subjects are reviewed, along with the entirety of each proposed study, by an institutional review board (IRB) charged with ensuring the ethical treatment of human subjects and their data. As a highly relevant example, data from the landmark “Six Cities” study that established a link between air pollution and human mortality, could not be made publicly available for a number of reasons, including promises that were made to the study participants at time of enrollment. Notably, the outcomes of this particular study have been validated in a number of scientifically rigorous ways and by several independent studies. Nowhere does the NPRM suggest that other mechanisms could be used to give the EPA confidence that the findings should be considered in rulemaking. Such mechanisms include comparing outcomes of several trials from different groups that reach the same conclusions, and vetting the science through expert scientific panels specially convened for this purpose. These mechanisms are especially important for studies conducted in the past, for which the underlying datasets may be wholly unsuitable or unavailable for public review.
In some instances, this proposed rule will directly conflict with Executive Order 135565 and the accompanying National Archives and Records Administration implementing directive, which place limitations on the release of specific non-classified information involving privacy, security, proprietary business interests, and law enforcement investigations. We would particularly question if controlled unclassified information (CUI) categories limiting the release of information relating to health, genetic information, proprietary business information, pesticide producers and railroad safety analysis records might prevent valid scientific studies based upon these CUI categories from being released and as a result, prevent the best possible scientific evidence from being used to develop regulations aimed at protecting public health and safety.
Even for those studies where the researchers are not specifically prohibited by an IRB or other restrictions from making the data publicly available, de-identification to maintain privacy is not simply a matter of redacting names from documents, but a complex and resource-intensive process. Scientific data may be prepared for and shared with other scientists for many reasons, including the development of new hypotheses, new analyses to seek novel patterns or test current suppositions, or for purposes of reproducing or confirming aspects of a particular study, and also to avoid duplication of efforts and accelerating discovery. Weighing the merit of these approaches, the privacy considerations, and the resources needed to accomplish meaningful data sharing is what drives these decisions today, and should continue to be considerations going forward. Policy should be informed by science, but science is not always undertaken for purposes of informing regulatory decisions.
Even the strongest and most sincere supporters of the open science movement have recognized that here is value in research for which underlying data are not made publicly available and acknowledge an imperative to leverage all science to develop policies and regulations. The goals of open science are not advanced through this proposal, which does not provide incentives, funding, or infrastructure for increasing access to data, but simply allows the Agency to disregard important, ethical, well-designed and executed studies.
The trend toward data transparency championed by the Associations and by foundations and science agencies around the world is predicated on the view that peer-reviewed, published studies already have presumptive scientific merit. In fact, the EPA itself is bound through a recent Executive Order to develop regulations through “transparent processes that employ the best available peer-reviewed science.” The proposed rule presumes that peer-reviewed, otherwise credible scientific studies do not merit consideration in the agency’s decision making, unless data supporting the study are made publicly available within the agency’s specifications. This presumption is not in keeping with the spirit of data transparency as understood by the scientific community. Science does not depend on the public availability of underlying data to indicate quality and reliability of evidence, and public availability of research data is not a proxy for the reproducibility of science. The Associations urge the EPA to promptly rescind this proposed rule, and to engage with the scientific community to discuss how evidence-based policy should be developed to protect human health and the environment.
Darrell G. Kirch
President and CEO AAMC
Mary Sue Coleman
Anthony P. DeCrappeo