From anecdote to evidence: Mapping the impact of US government action on public health research

October 27, 2026

Report finds threats at every level of the research system, from tampering with the research record to conditioning funding and policing what researchers can say, and sets out recommendations for protecting research from censorship.

A new report from #DefendResearch and Research Consulting, From anecdote to evidence, assesses US federal censorship of knowledge in public health research, an area where the consequences are especially high. It provides a citable evidence base on how these actions are affecting research.

Even when initial decisions were reversed, much of what was lost has not come back intact. Only about a third of the reversals the study tracked were straightforward restorations. Most were partial, conditional or reversible: some federal health data tools were put back online carrying notices that the agency intends to remove them again.

The study analyses more than 4,200 coded extracts from 244 scholarly and news sources describing individual incidents. Taken together, these incidents reveal a set of mechanisms operating across the research system that constrain what evidence can be produced, preserved, published, communicated and used for future research, clinical practice and policymaking.

The study identifies threats at three interconnected levels:

Across the research system: tampering with the research record. Data has been removed, altered or no longer collected. Gender identity and sexual orientation questions, for example, were dropped from roughly 360 federal data collections.

Within institutions: conditioning funding. Expert advice was overruled, funding cut and journals pressured. NIH alone saw more than 5,776 grants worth $14.7 billion disrupted.

For individual researchers: policing what researchers can say. Veterans Affairs scientists were ordered not to publish or speak publicly without clearance from political appointees, and evidence has been found of diversity language decreasing in use over time, a form of self-censorship that leaves little public trace.

Courts, philanthropists, journals and professional bodies have pushed back in three ways: reversing government actions, replacing lost resources and refusing to comply. But these responses have recovered assets without restoring control. Federal data, funding and scientific institutions still depend on decisions made by the administration.

The findings matter well beyond the US. Researchers around the world rely on US federal data and research infrastructure, and the pressure has reached international publishing: the US Health Secretary said he was considering barring government scientists from publishing in The Lancet, JAMA and NEJM. More than 5,300 USAID contracts were cancelled by early March 2025, disrupting global health programmes that other countries and donors have only partly replaced.

The report concludes with a preliminary set of recommendations intended to spark a wider, solutions-focused conversation about the changes needed to protect research from censorship. These include legislating, not only litigating, to protect public access to federal health data; treating alternative data systems as insurance, not a substitute; and giving shared legal and financial backing to institutions that refuse political demands. They offer lessons for any research system that wants to be ready before similar pressure arrives.