Briefing on Data Access

February 26, 1999

The University Community and the A-110 Proposed Revision

Mary Ellen Sheridan
University of Chicago

I feel that first I should give a disclaimer that as we move down the panel, there is less and less original content.

I have been asked to comment on the potential impact within the academic research community of Senator Shelby's amendment and OMB's proposed implementation of this statute. First, let me firmly state that the university scientific community is the strongest voice for data access, for the right to use data in an unfettered way, and for the public dissemination of new knowledge. From the standpoint of research administrators, we spend most of our time negotiating so that we ensure that privilege, and that we uphold that academic principle when we interact with industry, the federal government, or any private sponsor. Hence, the remarks of caution that I will give should not be equated with constructing barriers to effective public access to scientific research. However, the sweeping language of the statute fires grave concerns among our academic researchers about the need to balance the integrity of the scientific process and inquiry, as well as public accountability. The generation of new knowledge is a delicate but rigorous process. What we remember about the essence of science and how we learned the scientific method underlie the issues that we will raise about the integrity of the process. Because this is so critical to us, both as scientists and in our duty as educators to foster the public understanding of science, including measures that safeguard the integrity of the process, it may be useful to restate those fundamentals. The traditional steps of the scientific method are these: formulation of a hypothesis, development of rigorous experiments with controls and variables, collection of data, analysis of that data against the hypothesis, and a process of refinement and reformulation of the hypothesis. Scientific research is hard work. Collection of data is not a singular event; it is a process with checks and counterchecks, scrupulous attention to detail, and exacting measurements. It involves replication; it involves reproducibility. The interpretation of results is an evolutionary process. It does not happen just once and then disappear.

Let us not be confused. The storage of data, access to stored data, and understanding of that data are not fungible. Data alone are not science. Data in isolation, in fact, are fairly fragile. Premature data are even more fragile. Hence, the benefit of scientific inquiry is dependent on robust data. The dissemination of new knowledge is the vigorous defense of conclusions. Others have commented on the process by which scientists begin to discuss their data, the informal sharing of results among colleagues, posters and presentations, and professional meetings. Gordon conferences are an excellent example in which attendees sign a statement that they will not provide any public disclosure of the information. This is probably the best of opportunities for scientists to speculate on what is happening.

And science is a speculative process. I am concerned about the informal, but regimented, annual reports that are required by federal agencies of their grantees. Our scientists are more often than not using words such as "we have preliminary indications," "we have preliminary data that support our hypothesis," or "we have evidence that." If "publication" were interpreted to mean those windows to program officers and agencies of the excitement indicated by the process of science, we would have devastating outcomes. Probably the most devastating of them would be silence, because investigators would feel that they could not share with the agencies that had funded them what was happening in the laboratory; that it would be premature. And so, we are extremely concerned about how "publication" could or would be interpreted.

There is art in science. There are subjective judgments. Sometimes those become that leap of faith that garners medals, and sometimes they are based on sloppy or misguided thinking that leads to cold fusion fiascoes. The submission for publication is the reality check. That is the bedrock of academic science and biomedical research; the submission of information when scientists feel confident enough that they have the valid data, the appropriate conclusions, or when their preliminary tests have been satisfied in conversations with colleagues in Gordon conferences. The test of judgment of science is manifested in the peer vetting and publication process. This validates the methodology. It examines and interprets the data, gives verifications of reasonableness, and provides significance to the results. We are extremely concerned that the danger of early intervention into this scientific process is entirely possible under both this statute and the implementation, as well intentioned as they may be. Incomplete data clearly will be subject to error. Data point 17 may be of little consequence in the long run, and in longitudinal data, early indicators may not be borne out as valid evidence of conclusions that could be drawn. We are concerned that lack of expertise in properly interpreting data may lead to very damaging misinterpretations of science. Incomplete outcomes have led, and may continue to lead in the future, to public harm and public loss of confidence in the credibility of scientific research. We are also concerned that early intervention and interception of data could lead to mischief and allegations of scientific misconduct. I think the potential for the interruption of the research process and the dissemination of peer-reviewed science is the leading concern of the academic community.

We do have other concerns. There are collegial competitors within the academic community, within hospitals and biomedical research. Most are well behaved, and the chase is for the best quality and productivity of our laboratories. We fear, however, the unscrupulous people who would take advantage of another's insights, creativity in concept, methodology and hard work to leap frog to the head of the pack. It could happen among our academic communities. We are probably more gravely concerned about the opportunity for commercial organizations to both use FOIA, and to be taken advantage of by FOIA.

First of all, the private sector could use the FOIA tool for raiding basic research and scientific data that are being federally funded. We are concerned that the FOIA tool could sanction a new type of corporate welfare, franchising the university community for industrial outsourcing of basic research through federal dollars. Second, we are concerned that scientists have no recourse but to contain their federal research to the conservative and the mundane, and to take their most inventive and cutting-edge research outside federal reach. Not to their basement laboratories; that is probably unlikely to happen. Scientists cannot typically afford to do that. However, they may look for sources of support for the most leading-edge, creative research that will assure that they do not open their research to these FOIA interventions.

The other commercial-sector concern we have is the disruption of the partnerships that we have built with industry. Federal agencies, which advocate public/private/university partnerships, may find these irreversibly damaged if the private sector participants believed that corporate information that they have shared with and brought to these partnerships would be accessible through FOIA.

We believe there is potential harm to students. Graduate students would be working for years on dissertations, only to discover that the area of research on which they are working has become of interest. A public interest group or the private sector could reach through into that data, undermine the students' dissertations and cause them problems with respect to getting their dissertations published in peer-reviewed journals, which may claim that there has already been a public exposure of the information and that they would not be able to publish the dissertation.

I respect Senator Shelby's opinion in selecting FOIA as the mechanism for data sharing. Note, however, that the FOIA legislation and its exemptions were not designed to apply to the data held by grantees. The National Institutes of Health's implementation of FOIA was that grantee organizations are the intended possessors of the data generated with federal support, and in fact, their FOIA instructions exempt data held by grantees through FOIA. FOIA was intended to access records and other information in the hands of the federal agencies. Our concerns about the selection of FOIA as an instrument are first that FOIA intends to govern the actions of the agency. There is no opportunity for a grantee organization to ensure that the data are transferred to the public appropriately, or that promises of confidentiality given in good faith to human subjects, the private sector, or international collaborators would be honored by the agency. Under this statute, all of the decision-making authority for determining how and which data are shared is transferred from the university or the grantee organization to the federal agency, along with the raw data. Even a well-intentioned and honorable agency could find the exemption for medical records or privacy being challenged in its applicability to research data with dreadful consequences. The loser would not only be the biomedical researchers, but the public, as volunteers step away from participating in clinical studies and other informing and enriching research that is dependent upon human subjects. Just imagine yourself reading an informed consent form that would now be required to read, "This research is funded by the federal government. Therefore, information about your participation in this project may be available to requestors through the Freedom of Information Act."

What about data already in existence where confidentiality commitments would surely be breached in the mere transfer of the data to the federal agency? We have grave concerns about the reach-through to already existing data that were collected under assurances of confidentiality or assurances of respect for proprietary information that were given to our collaborators, whether they are human subjects or whether they are actual participants in the research project with us. The exemption for the protection of business and commercial information may be weak in its applicability to universities as nonprofit organizations. Yet, the intellectual property that is developed in the course of federally sponsored research does have commercial potential. Public dissemination through FOIA of preliminary data may jeopardize the protection of that intellectual property through patenting and, moreover, may damage the ultimate transfer of that technology for commercial development and the obvious subsequent public benefit.

We are concerned about the definition of "data." We must expand beyond the laundry list of types of data that NIH has to include all types of geophysical science data, astronomical data, and engineering data. The type of data that is available in scientific research is unbounded in the way in which information can be gathered and recorded. This is the way it exists in our laboratories, in our university offices, and in our researchers' offices. There are already policies and guidelines from the federal agencies that are the chief sponsors of basic research that give direction for accessibility of tools, research material, and technologies that reflect our sponsors' expectations of grantees. These guidelines are published, and they influence the transfer of materials and the sharing of data within those research communities. These are good examples of the way in which data sharing policies can and should be developed. The selection of FOIA is not the way to do that.

It is very common that journals have expectations of data access and data sharing. More and more journals are doing that. It is one way to ensure that there is an opportunity to access data and to vett it in different ways than perhaps the researchers themselves have interpreted it. We are concerned about the articulation of what is in the public's interest. We talk about public accessibility, the public good of data sharing; we are concerned that really it is the benefits of science that are in the public's best interest. The academic community is very strong and very consistent in its defense of free and open publication. We want data to be free and open. We want people to be able to see the data, but to see the data that are fit to be seen. Publication and peer-review journals are the benchmark in most instances, but very long-term research may need other markers. Continuously evolving longitudinal databases provide different questions that will need different answers to the appropriate way to share data.

Further, when scientific data are used for the underpinning of regulatory rulemaking that has great economic impact, other protections may be in the public interest, and they should apply. These are probably best governed by the governmental agencies and those who funded the research in the first place. The validation of results and the defense of interpretation are the sponsor's responsibilities, as it bases its rulemaking and new policies and studies on such scientific outcome.

Remember prior to the Bayh-Dole Act results were more commonly in the public domain. However, the societal benefits were substantially retarded because of that, and in recognizing that the grantee organization owned the data and could protect the intellectual property that is associated with that data and commercialize it, we were actually achieving a public good, and not necessarily becoming more secretive with data. Academic institutions are analyzing the commercial potential of data. Bayh-Dole alone did not bring that out. I think the biotechnical revolution raised the consciousness of many scientists as to the commercial potential of basic research. This raises interesting questions. The way we answer these questions must take into account the integrity of the scientific process and the public good.

The cost is not merely financial, for it is not merely a matter of photocopying laboratory notebooks or databases. The cost comes in making the data understandable in the documentation and in diverting scientists from productive science to transporting of raw data. Who will be responsible for transmitting data and educating the requestor when data is undecipherable in its raw form? How would sponsors - and I do not mean the federal government; many of these are partnerships - feel about how we had identified and appropriately protected the data accessible through FOIA?

We are very concerned about the diversion of the grantee resources away from the actual conduct of science to the presentation of data suitable for public review. This intrusion of a regulation into an orderly process of scientific productivity is extraordinarily costly. The effort required by our senior faculty and key investigators to appropriately respond to these FOIA requests will be enormous. I think it will be beyond the cost.

There may be some recourse. We appreciate that OMB, in reading the statute, listened to our concerns and that it attempted to formulate an implementation that addressed publication, rulemaking, and policy use. However, we are convinced, as time passes, that the best-intentioned implementation will not adequately protect us when we begin with bad statute. We believe that OMB should take the opportunity now to use its full powers of deliberation and turn to agencies and external bodies for additional guidance. This is not something that will be resolved in the short term. We know that there are opportunities and needs for data sharing policies to be developed. OMB could make adequate provision for grantee organizations to have a key participatory role in the determination of which data are turned over to federal agencies. It cannot be raw data that we have committed to keep confidential. OMB must find a way to link the grantee organizations and the federal sponsoring agencies if there is any mechanism for this. The link is not through FOIA. Thank you.