On Jan. 31, 2025, I was taking a break from an afternoon of meetings to check the news when I noticed multiple reports of datasets, info sheets, and other information resources disappearing from the Centers for Disease Control and Prevention and National Institutes of Health websites.

Pages dedicated to gender-based violence, LGBTQ+ youth, racism in health care, and vaccine recommendations had disappeared. Datasets relating to HIV and TB, along with the Youth Risk Behavior Surveillance System (a survey of health-risk behaviors in adolescents), were gone. The purge was not limited to the NIH and CDC; information and datasets from the Census Bureau, Department of Justice, the VA, and many other agencies and bureaus were impacted. By Feb. 2, The New York Times reported the removal of more than 8,000 federal web pages.

The removal was not entirely unexpected. In the lead-up to the presidential inauguration, librarians and others who regularly work with government data and information were preparing for the new Trump administration to minimize any federal efforts aimed at improving diversity and equity. During the weeks after the inauguration, the NIH and CDC were under a communications pause, which included the CDC’s signature publication Morbidity and Mortality Weekly Report, along with all communications from the National Library of Medicine. Although it appeared that PubMed—the largest biomedical literature database in the world—was continuing to add new content via automated workflows, no one from NLM was allowed to confirm this.

The American Library Association Code of Ethics includes this sentence: “We uphold the principles of intellectual freedom and resist all efforts to censor library resources.” For many of us, not only did the deletion of these resources suddenly make our work far more challenging, it cut directly to our professional values and ethics. At the Brown University Library, librarians in Health and Biomedical Library Services (HBLS) began to track reports of missing information and compiled an internal list of alternative or archived sites for medical and public health resources.

The data tracking and rescue efforts extended across the library. In late January, Frank Donnelly, head of GIS and Data Services at Brown, began to write Python programs to scrape and download some government datasets of concern, to ensure continued access. In early February he joined the Data Rescue Project, an international effort created to serve as a clearinghouse for data rescue-related efforts and data access points for at-risk public U.S. governmental data.

The urgency of the issue was clear, and later that month the library convened a Data Preservation group under Donnelly’s leadership. This was a cross-disciplinary team with members from CLEAR Data Services, HBLS, and the Center for Digital Scholarship, in recognition that these actions would affect anyone who worked with federal data or information resources. The charge for the group:

1. Stay up to date with archiving projects and new repositories for federal government datasets, as well as alternate sources for data.
2. Answer inquiries from students, faculty, and staff about finding missing datasets, and address concerns about the removal of data.
3. Actively contribute to the Data Rescue Project.

After months of legal action, the administration agreed to restore the impacted resources, generally with a disclaimer at the top of each webpage stating, “This page does not reflect reality and therefore the Administration and this Department reject it.” My colleagues and I are still concerned about stability and reliability of existing data and information—not to mention the confusion such a disclaimer could cause anyone seeking information. Our concerns generally fall into three categories:

1. Will additional (currently available) resources be removed because of their content?

   These “at-risk” resources are the focus of Brown’s Data Preservation group. So far, the group has downloaded and archived over a dozen datasets, including a large collection of USAID Demographic Health Survey data, bird flu data from the U.S. Geological Survey, and the Climate Change and Human Health Literature Portal from the National Institute of Environmental Health Sciences.

2. With datasets and databases that remain available, how do we assess validity?

   Although many of the datasets and information resources are back online, future iterations may not include questions about sexual orientation and gender identity, and information directed to specific population groups may no longer be updated. A recent Lancet study found that many governmental datasets have recently had descriptive terms altered without appropriately logging the change (e.g., changing “gender” to “sex”). MedlinePlus.gov—long touted as an unbiased source of health information—no longer contains a section for American Indian and Alaska Native health.

   Nicole Snyder Dettmar, a librarian at the University of Washington, recommends this evaluation strategy:

   • Look at the currently published information (try Population Groups on MedlinePlus).
   • Compare it to previous versions using archive.org.
   • Has terminology or content changed?

3. As agencies and institutes are being defunded, will critical resources be maintained?

   For medical librarians, clinical providers, and biomedical researchers, PubMed is the backbone of our work and is used daily to inform research and clinical practice. Many of us worry not just about the stability of the infrastructure given large-scale layoffs at the NIH, but also the integrity of both the content and the process for adding resources.

   PubMed is not the only at-risk site. The Education Resources Information Center briefly stopped updating in April due to staff reductions at the Department of Education. (It later was announced that ERIC would continue “with a more refined scope.”) International concerns over these resources are severe enough that the German National Library of Medicine announced plans to develop an “open, reliable, and sustainable” alternative to PubMed.

This is a quickly changing landscape; in July, news broke that hundreds of governmental websites, including vaccines.cdc.gov, will be decommissioned as part of a multi-agency review by the General Services Administration.

On Guard

The Brown University Library is responding to this moment by providing reliable information and data resources, educating students and practitioners on information evaluation, and participating in local and national data rescue and preservation efforts. Our guides to alternate sources for Federal Health Resources and to U.S. Government Data provide information about other ways to search PubMed and to find pre-2025 datasets. We subscribe to evidence-based clinical and research resources and purchase resources that provide archival coverage of government documents and federal publications.

I have spent my career as a medical and public health librarian, promoting the importance of critical thinking skills, information evaluation, and evidence-based practice. I also served from 2018 to 2022 on my local school committee and brought that same emphasis on best evidence to my public service. Earlier in my career, I regularly taught evidence-based practice and information management to MD and MPH students and to local public health workers. I was always proud to remind them that if the information came from the CDC or NLM, it was trustworthy.

As a citizen and as an information professional, I am saddened that I can no longer make this statement. The manipulation and censorship of information are contrary to the principles of free inquiry. It will have immediate harms to health care and to research, and will ultimately put us all at risk.