The Trump administration has deleted or modified at least 366 federal datasets across more than 60 agencies, according to a new tracker built by dataindex.us, a coalition of data policy experts, researchers, and former government officials. Twenty-eight datasets have been fully terminated; another 338 have been altered. The scope spans environment, labor, law enforcement, and education — but health data takes the heaviest hit, comprising roughly 40% of all affected collections. The mechanism is straightforward: datasets that produce findings misaligned with administration priorities are being quietly pulled, defunded, or stripped of questions the administration finds politically inconvenient. Multiple CDC surveys — including the National HIV Surveillance System, the National Violent Death Reporting System, and the National Intimate Partner and Sexual Violence Survey — have had questions about gender identity removed under Trump's executive order on "Defending Women from Gender Ideology Extremism." At least four health datasets have stopped updating entirely, including SAMHSA's Drug Abuse Warning Network, which tracked drug-related emergency room visits in near real-time to detect emerging overdose trends. The deletions extend well beyond health. The USDA's twice-annual survey of farm employers — which counted farmworkers and set the minimum pay rates for H-2A visa workers — has been discontinued. A National Center for Education Statistics survey tracking the pay and working conditions of teachers and principals at public and private schools is gone. A database of federal law enforcement officer misconduct records has been taken down. A survey measuring physical, sexual, and emotional violence against children and youth has been removed. What makes the damage hard to quantify is that there is no requirement for agencies to notify the public when they kill a data collection. Some datasets have simply stopped updating because staff were fired or reassigned. Others were pulled from websites with no announcement. The dataindex.us tracker was built precisely because this opacity is a feature, not a bug — crowdsourcing reports from media, government employees, and data users to identify what has vanished. "We're at a very vulnerable point as a nation," said Denice Ross, former US chief data scientist under Biden and cofounder of dataindex.us. "We're facing a future where we are not going to have the information that we need to make decisions that will save lives and benefit American communities." Thomas Babor, a professor emeritus at the University of Connecticut specializing in public health and addiction, put it more bluntly: killing these surveys is "like we're erasing our institutional memory for public health." The tracker itself is an admission that the full count is almost certainly higher. Christopher Dick, cofounder of dataindex.us and former chief of the Census Bureau's population evaluation branch, acknowledged that regular dataset updates can be hard to distinguish from permanent termination, and that obscure but critical datasets may not yet be on anyone's radar. The group is actively soliciting tips from people who depend on datasets that may have quietly disappeared. The pattern is not accidental. It is a deliberate strategy: if the data doesn't exist, the problem doesn't exist. No HIV surveillance by gender identity means no evidence of disproportionate impact on trans populations. No real-time overdose tracking means no early warning system for the next fentanyl wave. No farmworker wage surveys means no inconvenient baseline when setting visa worker pay. The administration is not arguing with the data — it is making the data cease to exist.
